Dissertation Data Analysis Guide for Computer Science Students by pwv16287

VIEWS: 0 PAGES: 62

More Info
									department of computer science


            at Duke university




                  Department GuiDe
department of computer science at duke university
                                                                                    department Guide


         welcome . . . . . . . . . . . . . . . . . . . . . . . 2

         proGrams of study . . . . . . . . . . . . . . . . . . . . . . . 5
         Graduate proGram . . . . . . . . . . . . . . . . 6
         underGraduate proGram . . . . . . . . . . . . . . . . . . . . . . . . . 7

         research projects . . . . . . . . . . . . . . 8
         Geometric computinG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
         internet systems, networkinG, and security . . . . . . . . . . . . . . . . 10
         memory systems and massive data manaGement . . . . . . . . . . 11
         BioloGical computinG . . . . . . . . . . . . . . . . . . . . . 12
         learninG and modelinG . . . . . . . . . . . . . . . . . . . . . . . . 13

         research collaBorations . . . . . . . . . . . . . . . . . . . . . . 14

         faculty . . . . . . . . . . . . . . . . . . . . . . 17
         alGorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
         artificial intelliGence . . . . . . . . . . . . . . . . . . 26
         scientific computinG . . . . . . . . . . . . . . . . . . 34
         systems and architecture . . . . . . . . . . . . . . . . . . . 40
         education . . . . . . . . . . . . . . . . . . 52
welcome
the department of Computer science at duke
university excels in research, teaching, and
learning in computer science, and engages with
the broader community at duke, in research
triangle Park, and beyond to impact progress in
computing and information technology. We are
a top-tier department that is at the forefront
of national research in many strategic areas
of computer science, that recruits and trains
                                                              Pankaj k. agarwal, Chair
the most talented students, and that provides
leadership in computing and information
technology. We achieve these goals by building
an excellent faculty, performing ground-breaking
research in core areas of computer science,
promoting multidisciplinary research, integrating
research into the curriculum, and forming strategic
partnerships. this guide gives you a glimpse
of what we are. Please visit our Web site at:
www.cs.duke.edu for more details.




2
welcome : department of computer science at duke university
computer science at duke university

One of the many hallmarks of our success
has been fruitful collaborations among
different groups within the department,
with research groups in other departments
at duke, with other institutes, and with
industry. as we move into the information
age, the focus of science is shifting          and education. the department is using a       being conducted, sponsored by various
from the discovery of new information          dual approach to combine research and          funding agencies.
to the computationally intensive task of       education. Bringing research into the             the department provides an extremely
processing and analyzing information.          curriculum is the best way to train students   stimulating, productive, and friendly
  We have outstanding programs in              about the most advanced technology and         environment in the form of classroom,
geometric computing; internet systems,         to disseminate the latest developments of      office, and lab space; computing
networking, and security; biological           computing technology in society.               infrastructure; teaching support; and
computing; memory systems and massive            We encourage undergraduate students          graduate fellowships and assistantships.
data management; and learning and              to get involved with ongoing major             it enables faculty and students to
modeling. the research interests of our        research projects through undergraduate        accomplish their full potential. the
faculty overlap with these areas and with      theses, research experience for                department is constructed to encourage
research areas in other disciplines such as    undergraduates (reu) support,                  innovative collaborations among the
biology, engineering, nanotechnology, and      independent studies, etc. some of our          sciences, engineering, environmental
environmental sciences.                        exceptional first majors graduate with         studies, and medicine.
  We also do work in a number of other         distinction, which involves a significant
important areas including computer             research component, and in many cases
graphics and visualization, sensor networks,   the research has resulted in publications
numerical analysis, software engineering,      in leading conferences.
complexity theory, and robotics.                 the eminence of our research and
  the department is arguably unique in         teaching faculty is the greatest strength
the symbiosis that exists between the          of the department. Many faculty members
education group and the research faculty.      have been recognized both at university
the synergy between them has been a key        and national levels for their excellence in
to the success in continually reforming        research, education, and service. Highly
the curriculum and integrating research        visible, multidisciplinary projects are


                                                                                                                                                                                               3
                                                                                                                                   department of computer science at duke university   : welcome
duke university

duke university is recognized as one of
the leading educational and research
institutions in the united states. Founded
by James Buchanan duke in 1924, its roots
can be traced back to 1838. the Graduate
school was established in 1926. the
university features top-ranked Medical,
Business, Law, and divinity schools.
the university enrolls approximately
6,500 undergraduate students and 7,117
graduate and professional students.
  the university is located on a beautiful
1,635-acre campus in durham, north
Carolina, a city of approximately 150,000
inhabitants. durham is at the apex of
north Carolina’s famous research triangle,
which includes the nearby university of
north Carolina at Chapel Hill and north
Carolina state university in raleigh. the
adjacent research triangle Park, situated
in 6,750 acres of rolling woodland, is
home to many sophisticated research and
manufacturing facilities. duke is centrally
located for a variety of cultural activities,
and the mild climate makes the area a
sports paradise. there are outstanding
recreational opportunities in the Outer
Banks (the eastern end of the state) and
the Blue ridge and Great smoky Mountain
ranges (in the western part of the state).



4
welcome : department of computer science at duke university
proGrams of study                                     a p p ly i n G t o t h e                        a p p ly i n G t o t h e
                                                      Graduate proGram                                underGraduate proGram
the department of Computer science at duke            admission to the department                     information about the
                                                      of Computer science is highly                   program is available at
university offers innovative research and teaching    competitive. information about                  cs.duke.edu/education/undergrad.
programs leading to degrees of doctor of Philosophy   the program is available at                     For information and the
(Ph.d.), Master of science, (M.s.), Bachelor of       cs.duke.edu/education/graduate.                 on-line application form,
                                                      For information about the Graduate              go to admissions.duke.edu.
science (B.s.), and Bachelor of arts (B.a.).          school at duke university, and                  special inquiries may be
                                                      the on-line application form,                   directed to dus@cs.duke.edu.
                                                      go to gradschool.duke.edu.
Our degree programs:                                  special inquiries regarding the                 financial aid for
+ provide instruction leading to a broad              admission process may be directed               underGraduate students
  intellectual understanding of the basic             to admissions@cs.duke.edu.
                                                                                                      undergraduate students may receive
  areas of computer science                           financial aid for
                                                                                                      financial aid through need-based aid,
                                                                                                      merit scholarships, student employment,
+ teach skills in the use of the tools and            Graduate students
                                                                                                      and loans. duke university admits u.s.
  technology of computer science                      Full-time Ph.d. students are awarded            citizens, permanent residents, and a
+ lay a firm foundation that students                 tuition and fees as well as stipend support     limited number of foreign students,
                                                      during their time of study in the form          without regard to financial circumstance
  can adapt to future technological change            of fellowships, teaching assistantships,        or aid eligibility and meets 100 percent of
                                                      and research assistantships. Financial          each admitted student’s demonstrated
                                                      support is not available for the Master of      need throughout nine semesters of
                                                      science program. the Graduate Program           potential undergraduate enrollment.
                                                      Coordinator assists students in need of         For more information on various aspects
                                                      financial support to identify potential         of the financial aid process go to
                                                      funding sources.                                dukefinancialaid.duke.edu.




                                                                                                                                                        5
                                                                                  department of computer science at duke university   : proGrams of study
d o c t o r o f p h i l o s o p h y ( p h . d. ) p r o G r a m                                   master of science (m.s.) proGram

the Ph.d. program encourages students                                                            the M.s. program combines a firm
to initiate research work early in their                                                         grounding in theoretical foundations
graduate study. students familiarize                                                             with training in current technologies
themselves with the faculty and research                                                         and applications. the program also
being conducted, attend research group                                                           incorporates research experience and
seminars and colloquia to expose them                                                            depth in a chosen area of concentration.
to a range of topics and problems that                                                           the M.s. degree requires 30 course
may be chosen for research projects,                 of graduate study as a Preliminary          credits with at least eight advanced
and enroll in classes to fulfill the course          examination. the Ph.d. candidate then       courses in computer science or a related
requirements for the degree. during the              conducts research that culminates in the    field, chosen to satisfy breadth and
first two years students complete the                doctoral dissertation defense.              depth requirements. M.s. candidates
general course requirements, conduct                   the anticipated time to earn the doctor   also complete a research project or
an initial research project under                    of Philosophy degree is five years.         thesis under the supervision of a faculty
faculty supervision, and gain at least                                                           member. M.s. students participate
one semester of teaching experience                                                              in a special research seminar during
as a teaching assistant. a breadth                                                               their first semester to expose them to
requirement is fulfilled by passing a                                                            computer science research. the M.s.
qualifying exam or receiving a “quals                                                            program provides flexibility for part-time
pass” for coursework in four of six core                                                         enrollment and is considered “terminal”
areas of subject knowledge. the initial                                                          in that it does not normally advance into
research project experience introduces                                                           the Ph.d. program.
students to research methodologies,                                                                the anticipated time to earn the Master
the current state of the art in the area                                                         of science degree is two years.
of concentration, and presentation of
research results. Having successfully
completed the course, breadth, and
research requirements during the first
two years, a Ph.d. student continues
independent research and prepares a
research proposal for the dissertation,
which is presented in the third year


6
proGrams of study : department of computer science at duke university
underGraduate proGram

Computer science embraces both the                                                           networks, databases, scientific computing,         the department has several national
science and art of processing information                                                    artificial intelligence, graphics,               science Foundation projects specifically
using computers. it combines the abstrac-                                                    computational geometry, computational            targeted at including undergraduates
tions and elegance of mathematics with                                                       biology, distributed computing, theory           in sponsored research. undergraduates
the scientific and practical aspects of                                                      of computation, and digital logic. the           have been coauthors on papers in major
engineering. students in the undergraduate                                                   department also offers interdisciplinary         conferences in many areas.
computer science program at duke build a                                                     minors for students who wish to pursue              the student chapter of the association
strong foundation in complex systems and       computer science, or who are interested in    computational aspects of social or natural       for Computing Machinery (aCM) has an
algorithms used in a variety of specialties.   a more rigorous and disciplined approach      sciences, such as economics or biology,          active series of talks and helps organize
they are prepared for continuing study         to the major, often choose the B.s. degree.   without majoring in computer science.            the regional student programming
or work in such wide-ranging fields as           the department also supports a                several programs outside the classroom         contest at duke. students interested
the internet and Web-related services,         cooperative B.s. double major with the        strengthen the undergraduate course              in participating in the contest are
bioinformatics, artificial intelligence,       department of electrical and Computer         of study. undergraduates interested in           encouraged to take a problem-solving
algorithmic and theoretical aspects of         engineering. students in both majors          participating in research are encouraged         seminar each fall that helps prepare for
computer science, and data analysis in         have had great success pursuing technical     to do so and are actively involved in nearly     the contest. a duke team has advanced
scientific and commercial applications.        careers related to computer science           all faculty research. Junior and senior          to the world finals in nine out of the
  duke offers several courses of study for     including consulting, programming,            undergraduates are encouraged to pursue          last thirteen years, placing as high as
undergraduates interested in computer          financial, scientific, and artistic jobs      independent study courses in areas they          third in the world. in addition, an aCM
science. students pursuing a B.a. degree       where an expertise in the area of             find intriguing and challenging. these           Committee on Women in Computing (aCM-
typically take courses toward a double         computer and information science is           courses are used to study an area in-            W) was recently established to celebrate,
major, minor, or certificate in another        expected and appreciated.                     depth, to develop expertise in areas             inform and support women in computing.
discipline. students have combined the            Both majors and the minor require a        not covered in the standard curriculum,          undergraduates also participate in the
B.a. degree in computer science with           solid background in programming and           and to pursue research projects during           uta (undergraduate teaching assistant)
majors in Mathematics, economics, english,     the basic theoretical and practical issues    the academic year. the competitive               program which provides support for both
art, Biology, and nearly every major, minor,   that are part of the relationship between     C-surF (Computer science undergraduate           major and non-major courses including
and certificate program at duke.               programming and computer science. this        research Fellows) Program involves               staffing labs and help sessions, developing
  the B.s. degree combines a more              foundation is strengthened with required      undergraduates in an intensive, multiple-        assignments, grading, and mentoring.
in-depth and wide-ranging view of              100-level courses in architecture, systems,   semester research experience in a                undergraduates also participate in several
computer science with both mathematical        algorithms, and software design. electives    computer science or interdisciplinary            student organizations including duLuG
foundations and applications. students         and advanced courses are available in         project leveraging core concepts in              (duke university Linux user’s Group) and
intending to continue the study of             many areas of computer science including      collaborative ways.                              MuG (Microsoft user’s Group).


                                                                                                                                                                                               7
                                                                                                                         department of computer science at duke university   : proGrams of study
research projects
the colored chart represents our research
areas and projects. each faculty member
may be involved in one or more research
projects. Colored squares on faculty pages




                                                                                alGorithms

                                                                                             artificial intelliGence

                                                                                                                       scientific computinG

                                                                                                                                              systems and architecture

                                                                                                                                                                         education
differentiate on which project(s) specific
faculty members are currently working.




                                                Geometric computinG

                                                                                                                                                                                     internet systems, networkinG, and security

             memory systems and massive data manaGement

                                                                                                                                                                                                           BioloGical computinG

                                                        learninG and modelinG




8
research projects : department of computer science at duke university
research projects >   Geometric computinG

We are exploring issues such as how           at duke. they collaborate with faculty        another essential property, which we
techniques for solving problems can be        in Mathematics, Biology, Biochemistry,        usually take for granted, is that highways
discovered, improved, analyzed, and           electrical and Computer engineering,          are continuous, indeed connecting
demonstrated to be correct or optimal.        and the nicholas school of environment.       a to B, and not just approximately. the
We expect to make leading contributions       Beyond duke, the group also collaborates      corresponding subfield of computational
in computational geometry, geometric          with researchers at various top institutes.   geometry is often referred to as
modeling, data structures, high-              Because of its depth and breadth, the         computational topology. a good portion
performance computing, i/O-efficiency         geometric computing group at duke is          of our efforts may be classified to
for external memory, geographic               arguably the top geometric computing          belong to this subfield and are driven by
information systems (Gis), biological         group in the nation.                          applications in a variety of other fields,
computing, and data compression.                Geometric computing research at duke        the prediction of the structure of folded
  Because of the geometric nature of          works under the common rationalization        proteins and the reconstruction of human
the physical world in which we live,          of the field of computational geometry,       organs being two. the hallmark of our
geometric problems arise in any area          often given in the past, that the world       work is fast algorithms that implement
that interacts with the physical world.       around us is three-dimensional and            mathematical models to offer insights
Geometric computing focuses on designing,     questions how things in this world relate     into and answers to such questions.
analyzing, and implementing efficient         to each other are inherently geometric.
algorithms for geometric problems.            take moving a piano through a door-frame
the geometric computing group of the          and planning a flight-path that avoids
department is internationally renowned        collisions with other airplanes as two
for its contributions to the fundamental      examples. it should therefore not surprise
problems in computational geometry,           that computing properties about these
addressing massive data management            geometric things and their interaction
issues in geometric problems, and applying    are common-place and important. While
geometric techniques to a variety of areas,   this is still a valid argument, we would
including molecular biology, geometric        like to amend that most of the geometric
modeling, robotics, geographic information    questions people concern themselves
systems, ecology, and photonics.              with have to do with how things are
  the group actively collaborates with        connected. it is important that highways
other groups within the department and        are sufficiently straight and smooth
with the researchers in other departments     to support the driving of fast cars, but


                                                                                                                                                                                                9
                                                                                                                          department of computer science at duke university :   research projects
research projects >    internet systems, networkinG, and security

We are researching computer and internet              Our faculty members have established       techniques to detect programming errors,
technologies that are transforming the ways         a reputation for carrying out some of the    we are also developing techniques to
we live, learn, work, and play. advances            leading experimental systems work in         detect and recover from hardware errors
in these technologies provide the core of           the country. the goal of the group is to     (e.g., transient faults and manufacturing
the infrastructure that is driving economic         work on high-impact problems related         defects). in collaboration with operating
growth. the systems and architecture                to providing users with computing and        systems researchers, we are exploring
research group is currently working on high-        networking support for their applications.   techniques to manage energy as a first
impact problems related to providing users          the area is broadly defined, covering        class resource.
with advanced networking, distributed               topics ranging from internet-based             the group has established collaborations
internet systems, and pervasive wireless            services to ad hoc wireless networks,        with leading industry and academic
computing as the internet continues to              from massive data storage centers to         institutions and much of the ongoing work
permeate our daily lives.                           tiny sensor nodes, from the dataflow         focuses on internet operating systems
  in the 1990s, the public internet                 within hardware components to global         that improve robustness and survivability
emerged as a common basis for infor-                replication of databases across the          by adapting to the inherent dynamism of
mation sharing, large-scale computation,            wide area, and from security to energy       the internet environment, which features
communication, education, and                       efficiency. recent research projects         massive load swings, network instabilities,
entertainment. advances in network                  span architectures and operating             and unexpected failures and attacks.
technology have triggered a shift toward            systems for high-performance servers         these systems instantiate services
a service-based model of computing; in              and clusters, systems for wireless sensor    wherever adequate resources exist,
this model interconnected servers host              networks, and power management for           rather than binding them to a specific
shared applications and data repositories,          mobile computing devices, as well as         location. this direction aligns well with
accessed from client devices that are               fundamental techniques for building          long-term initiatives in industry and
increasingly specialized, portable, and             and evaluating robust, scalable, wide-       government—including major industrial
pervasive. this introduces tremendous               area distributed services. We are            partners—toward secure “grid” or “utility”
new opportunities, as well as new                   also exploring the performance and           computing for large-scale computational
vulnerabilities and fundamental new                 correctness of programs executing            science and commercial services.
architectural challenges for networked              on multithreaded systems (including
computing systems. as the scale of                  multiprocessors) by continuously
deployed systems increases, the focus in            monitoring programs for anomalous
this area will become increasingly network-         behavior (e.g., deadlock or excessive
centric and shaped by security concerns.            synchronization overhead). along with


10
research projects : department of computer science at duke university
research projects >   memory systems and massive data manaGement

a common theme in the research of many        and for restructuring program code to        in the size and format of data raises a
faculty in the department is memory           improve memory performance. We are           number of challenges in this area. in one
systems and massive data management.          exploring a variety of techniques to         of the projects we are designing derived
With advances in technology, massive          address fundamental issues that arise        data to handle the rich structure of XML,
amounts of data are becoming readily          from continually changing technology and     which is an emerging standard for data
available at a relatively low cost.           workloads. Our research efforts include      exchange over the internet and consists of
For example, experiments on Cern’s            high performance microprocessors,            semistructured graphs. in another project,
accelerators, satellite images from nasa,     multithreaded systems, nanoarchitecture,     we are developing efficient indexing
Ct-scans and Mri images of brains,            dependable computing, and energy             schemes for multidimensional spatial data
and signals from vast sensor arrays are       efficient computing.                         and for multidimensional data streams.
generating terabytes to petabytes of            at the algorithmic level, the goal is to     the group has been engaged in several
data every day. the current technology        develop algorithms and data structures:      multidisciplinary collaborative projects.
is not adequate to cope with such large       (i) that are specifically optimized for      in collaborative projects with the Pratt
amounts of data. expertise in a wide          memory hierarchy efficiency and that         school of engineering, it is experimenting
range of topics including algorithmic         minimize the access to secondary memory,     with high density nanoscale architectures
techniques, database systems, computer        (ii) that can handle long streams of         that will enhance the capability to access
architecture, distributed computing, and      data in real time, and (iii) that can        massive data and improve performance in
networking is needed to address these         provide tradeoff between efficiency and      high computational demand environments.
issues. the memory systems and the data       accuracy. i/O-efficiency, approximation,     a recent collaborative effort with the
management group in the department is         dynamization, and randomization are          Medical school is geared toward developing
unique because of its vertically integrated   some of the recurring themes in our work.    tools for integrating data spaces for
approach in addressing these issues—            the research in database management        medical and clinical research. another
from an architectural point of view, to       is broadly interested in data management     project in collaboration with the school of
an algorithmic point of view, to a            systems and their applications. Our          environment focuses on terrain modeling
database management point of view.            recent research focuses on maintaining       and ecological forecasting.
  at the architectural level, the goals       derived data, which is used to facilitate
are: (i) explore new processor and            the access of the base data. derived
memory architectures to improve overall       data is obtained by applying structural or
performance by addressing memory-             computational transformations to base
system bottlenecks, and (ii) develop          data, and it must be updated whenever
techniques for laying out data in memory      base data is updated. explosive growth


                                                                                                                                                                                           11
                                                                                                                       department of computer science at duke university   : research projects
research projects >    BioloGical computinG

We are striving to solve many complex               amount of work is being done on the          in collaboration with researchers from        nanofabrication of silver nanowires and
biological problems that seemed                     automatic classification of tissue types     other universities, the group is exploring    arrays of gold nanoparticles. We have shown
unapproachable until recently, and also             and automatic selection of relevant gene     dna self-assemblies of nano-electronic        the nanowires to be highly conductive and
taking a closer look at the possibility             features from gene expression data.          devices as a scaffold for nano-scale          are working to demonstrate single-electron
of developing alternative models of                    in collaboration with researchers in      computational devices that could one          transistors with the nanoparticles.
computation. Projects fuse biotechnology            computer science and biochemistry            day meet the semiconductor industry
and information technology— two of the              at duke and other universities we are        association’s grand challenge of
most significant developments of the                developing geometric techniques and          replacing conventional CMOs devices.
information age.                                    paradigms for representing, storing,         nanoscale computers have the potential
  On the one hand, advances in computer             searching, analyzing, and visualizing        to revolutionize computing similar to
science have played a critical role in              biological structures. One of the projects   the transition from vacuum tubes to
addressing and solving problems in                  in this area focuses on studying protein-    integrated circuits.
molecular biology, including the decoding           protein interaction. We have defined the       the dna nanotech group at duke is
of the human genome. the application                notion of interface surface between two      interested in understanding and exploiting
of new computational techniques to                  or more complexed proteins and have          the material properties of biomolecules
process, analyze, and visualize emerging            developed an efficient algorithm for         (mostly, dna and proteins) for applications
genomic data offers hope for solving many           computing the interface surface. Following   in dna computation, molecular databases,
complex biological problems that seemed             the ideas from Morse theory and topology     and nanofabrication. We are devising
unapproachable until a few years ago. On            we have defined elevation of points on a     methods to encode data and algorithms
the other hand, advances in biotechnology           molecular surface. We use elevation to       into 1d, 2d, and 3d structures of dna for
have raised the possibility of developing           identify features on a molecular surface     execution of molecular computations and
alternative models of computation.                  and to dock two proteins.                    wet database searches. We are engaged in
  the current strengths of the department             the area of biomolecular computation       the design, implementation, and analysis
in computational biology lie in functional          is an area of great potential because        of self-assembling dna nanostructures
genomics, analysis of protein structures,           of the massive parallelism available         with complex and specific 3d structures.
and biomolecular computation. in the                at the molecular scale. the duke dna         these structures are capable of scaffolding
area of functional genomics, research               nanotech group has demonstrated the          or templating other materials with diverse
is focused on the development of                    use of dna as smart glue to bring together   functional properties to form micrometer
principled methods for discovering                  electronically active nano-materials and     sized objects with nanometer-scale
genetic regulatory networks from diverse            also as templates for the formation of       feature resolution. For example, we have
types of data. in addition, a significant           highly-conductive metallic nanowires.        used dna self-assembly for “bottom-up”


12
research projects : department of computer science at duke university
research projects >   learninG and modelinG

We develop mathematical concepts                reliable information from the huge           industrial partner saiC to produce               study of skin lesions in the diagnosis
and computational methods in areas              quantity of imperfect data acquired          maps of physical environments with               of melanoma.
such as perception, reasoning, learning,        through sequencing studies, gene-            unprecedented accuracy.                            another project focuses on the
and planning. We investigate the                expression experiments, and from other         in human-computer interaction,                 development of a theory of stochastic
data structures and algorithms for              sources. these techniques combine            advanced models are at the basis of the          estimation for tracking the boundaries of
representing, reasoning with, and learning      stochastic methods with machine learning     “missing axiom theory” of dialogue: this         objects whose topology can change over
from visual, auditory, and other modes          and numerical optimization techniques.       theory proposes that participants in a           time. examples of these are living cells
of input. the recent, rapid proliferation         in collaboration with the school of        dialogue exchange the information needed         in the field of view of a microscope, the
of quantitative models in the natural           Medicine, the scientific computing group     to achieve their respective goals, and           consecutive cross-sections of a three-
and social sciences and in medicine has         has developed techniques for detailed and    interact to supply the missing axioms in         dimensional tomography scan of a human
underscored the importance of techniques        computationally efficient models of the      the proof that the goal has been satisfied.      body, or a cloud of pollutants spreading
for constructing models of complex              human heart and brain, of the chemical       this approach has been applied to man-           in the ocean or in the atmosphere.
systems and of algorithms for the efficient     behavior of different types of stimulants,   machine dialogue systems for equipment
manipulation of these models.                   and of the structure of molecules and        repair problems, mystery solving, and
  stochastic methods and machine                proteins. together with researchers in       automated tutoring. in recent work, the
learning techniques have proven to be           the school of engineering, the scientific    ai group has built a series of intelligent
powerful tools for tackling the increasing      computing group is also addressing the       automatic telephone answering systems
complexity of these models, and for             computational challenges raised by large     that make appointments or provide
harnessing the uncertainty inherent             sensor networks, and is participating        directory service.
in both models and measurements. in             in projects on compressive sensing             Computer vision research at duke
scientific computing, new hierarchical and      and on tracking with optical devices,        focuses on machine learning methods for
iterative techniques have multiplied the        bringing both classical and probabilistic    the analysis of shape and motion in both
opportunities offered by the rapid increase     techniques to bear on these problems.        images and video. For example, machine
in the power of today’s computers.                the artificial intelligence faculty        learning and linguistic analysis methods
  Our faculty are leaders in these and          have developed algorithms for solving        underlie solutions for the recognition of
related areas. their expertise ranges           stochastic control problems (MdPs) and       american sign Language, finger spelling,
from scientific computing and machine           robot mapping, as well as addressing         and gesture languages for interacting with
learning to artificial intelligence, computer   issues in man-machine interfaces. in the     computers and devices. Methods for the
vision, and computational biology. For          robotic mapping domain, an algorithm         automatic recognition of shape and for
example, functional genomics is benefiting      developed by the ai group has been           the detection of their changes over time
tremendously from techniques that extract       deployed on mobile robots donated by         are being employed for the dermatological


                                                                                                                                                                                             13
                                                                                                                         department of computer science at duke university   : research projects
research collaBorations                                                       Genome science                              environmental science


the department has many opportunities to engage                               recognizing the importance of genome        the goal of the Center on Global Change
                                                                              science and building on the strong          at duke is to develop the scientific
in multidisciplinary research and education. We are                           Medical school, at the end of the last      and technical basis for recognizing
contributing meaningfully to all of duke’s important                          century duke established the institute      and predicting changes in the earth’s
strategic initiatives. a few of the initiatives include:                      for Genome sciences and Policy (iGsP).      environment. the department is
                                                                              the department is actively participating    participating through collaborative
                                                                              in this initiative through the Center for   projects in ecological forecasting and
                                                                              Bioinformatics and Computational Biology    geographic information systems (Gis),
                                                                              (CB2),and the Center for systems Biology.   joint grants (e.g. ess and Bdei), and
                                                                                alexander Hartemink, Bruce donald,        teaching relevant courses (e.g. Gis,
                                                                              Pankaj agarwal, Herbert edelsbrunner,       ecological forecasting, and algorithms).
                                                                              and John reif are participating in iGsP.    Pankaj agarwal has been collaborating on
                                                                              their involvement includes conducting       developing complex forest growth models
                                                                              research in computational biology, co-      for forecasting ecological phenomena.
                                                                              advising Ph.d. students, and teaching
                                                                              some of the core and elective courses of
                                                                              the recently established Ph.d. program in
                                                                              Bioinformatics and Genome technology
                                                                              (BGt). in addition, the department offers
                                                                              a number of courses that are relevant to
                                                                              students in the BGt program. as the iGsP
                                                                              evolves, the department continues to play
                                                                              a key role in the program.




14
research collaBorations : department of computer science at duke university
photonics and                                 BioloGical and
communication                                 nano computinG
the Fitzpatrick Center for Photonics          the department is playing a leading
and Communication systems at duke             role in the duke nanoscience initiative
investigates common themes in advanced        through collaborations, initiation of
optical communication networks and            the nanoscience certificate program
the visual information encoded and            and nanoscience seminar series, co-
transferred on optical fields. its research   advising of graduate students, joint
programs include quantum optics, opto-        grants and grant applications, and
electronics, optical networks, information    shared equipment in the shared Materials
spaces, and biophotonics. the current         instrumentation Facility (sMiF). Members
collaboration between the department          of the department of Computer science,
and the Fitzpatrick Center focuses on         including John reif, Bruce donald, alvin
the research in information spaces.           Lebeck, and thomas LaBean, have active
the main challenge here is to meet the        research programs in various topics of
enormous computing, networking, and           nanoscience including experimental
storage needs of the data generated by        and theoretical work in self-assembled
massive sensor networks. the faculty          dna nanostructures, nanoelectronic
members in the geometric and scientific       architectures, and molecular robotics.
computing groups, including Xiaobai           the research in these areas of nanoscience
sun, are collaborating with the Center        is highly interdisciplinary, spanning not
on developing numerical and geometric         only the traditional computer science
techniques to design large sensor             disciplines of combinatorial design,
networks, to analyze their performance,       computer architectures, and robotics,
and to process the data.                      but also biochemistry, chemistry,
                                              physics, electrical engineering, and
                                              material science.




                                                                                                                                                                     15
                                                                                           department of computer science at duke university   : research collaBorations
duke computer science faculty
the department of Computer science has five main faculty groups: algorithms,
artificial intelligence (ai), scientific Computing, systems and architecture, and
Computer science education. innovative new research programs in geometric
computing; biological computing; internet systems, networking, and security;
memory systems; data management; and modeling, learning, and data analysis
result in complementary cross-group collaborations. Our faculty members are
among the best researchers in their fields. they provide valuable opportunities for
students to engage in stimulating discussions, exciting research, and challenging
problems. students and faculty explore innovative solutions involving both
theoretical and practical aspects of computer science. Our excellent faculty/
student ratio fosters close, collaborative relationships.




                                                                                      17
alGorithms
faculty
                                                                                                                                      alGorithms

                                                                                                                                                   artificial intelliGence

                                                                                                                                                                             scientific computinG

                                                                                                                                                                                                    systems and architecture

                                                                                                                                                                                                                               education
                                               research projects                                       Geometric computinG
                                the colored chart enables you to easily                                                                                                                                                                    internet systems, networkinG, and security
                                 identify on which research projects our
                                                                              memory systems and massive data manaGement
                                  algorithms group is currently working.
                            each faculty page displays a colored square                                                                                                                                                                                          BioloGical computinG


                              that corresponds to the research project.                                      learninG and modelinG




the field of algorithms (also known as foundations or theoretical computer science) is concerned with the
mathematical basis of computing. it explores issues such as how algorithms (i.e., techniques for solving problems)
can be discovered, improved, analyzed, and demonstrated to be correct or optimal. the field is strongly influenced
by practical concerns such as execution time, storage space, communication, and the constraints imposed by
hardware architectures.
  the algorithms group, rated among the best nationally, makes leading contributions in computational geometry,
geometric modeling, data structures, approximation, online, and combinatorial algorithms, algorithms for database
and data stream systems, high-performance computing, i/O-efficiency for external memory, geographic information
systems (Gis), and biological computing.

                             primary                                       secondary
                             + Pankaj k. agarwal                           + terrence Furey
                               (secondary appointment                        (primary appointment in
                               in Mathematics)                               Biostatistics and Bioinformatics)
                             + Herbert edelsbrunner                        + John Harer
                               (secondary appointment                        (primary appointment in Mathematics)
                               in Mathematics)                             + Mauro Maggioni
                             + thomas H. LaBean                              (primary appointment in Mathematics)
                               (secondary appointments in
                               Chemistry and Biomedical engineering)
                             + kamesh Munagala
                                                                                                                                                                                                                                                                                        19
                             + John H. reif
                                                                                                                     dePartMent OF COMPuter sCienCe at duke university :                                                                                alGorithms faculty
Geometric computinG


BioloGical computinG


learninG and modelinG




                                                       education:                                   Professor Pankaj agarwal’s research            (Gis) project, Professor agarwal studies
                                                       Ph.d., Courant institute of                  focuses on developing efficient algorithms     the problems of terrain modeling, terrain
                                                       Mathematical sciences, 1989                  and data structures for large-scale            navigation, tracking and searching moving
                                                       M.s., university of California,              geometric problems that arise in molecular     objects — a problem central to location-
                                                       santa Barbara, 1986                          biology, global change, databases, and         based services and military applications.
                                                       B.e., university of roorkee, 1982            mobile computing. By gaining geometric
                                                                                                    insights into these problems, he makes         selected publications:
                                                       research interests:                          them tractable on modern computer              + agarwal, P.k., kaplan, H., and sharir, M.
         pankaj k. aGarwal                             Computational and combinatorial              architectures. “i’m interested in anything       “Computing the volume of the union of
         rJr nabisco Professor                         geometry, computational biology,             that’s geometric, from the molecular             cubes.” in 23rd Annual Symp. on Comput.
         of Computer science                           robotics, spatial databases, geographic      level to the global level,” says Professor       Geom., 2007.
         Professor of Mathematics                      molecular information systems, and           agarwal. His mathematical models               + agarwal, P.k., edelsbrunner, H., Harer,
         department Chair                              data structures.                             employ approximation and randomization           J., and Wang, y. “extreme elevation on
                                                                                                    techniques to find simple, fast solutions.       a 2-manifold.” Discrete Comput. Geom.,
                                                       honors and awards:                             Professor agarwal is working on a              36, 2006.
                                                       aCM Fellow, 2002; Bass society of Fellows,   number of collaborations with many             + agarwal, P.k., Xie, J., yang, J., and yu, H.
                                                       2000; alfred P. sloan Fellow, 1996;          departments at duke university, including        “scalable continuous query processing
                                                       national young investigator, 1993.           the school of the environment, Biology,          by tracking hotspots.” in Annual IEEE
                                                                                                    Biochemistry, Math, and engineering. as          International Conference on Very Large
                                                                                                    part of a computational biology project          Databases, 2006.
                                                                                                    with Professor Herbert edelsbrunner, he
                                                                                                    develops algorithms to model proteins
                                                                                                    as geometric shapes to determine how
                                                                                                    protein structures affect these functions.
                                                                                                    this research can help doctors understand
                                                                                                    the biological mechanisms of illness. in a
                                                                                                    project with the school of environment and
                                                                                                    Biology, Professor agarwal is designing
                                                                                                    algorithms that forecast the long-term
                                                                                                    effects of deforestation on the population
                                                                                                    of various natural plant and animal
                                                                                                    habitats in regions of the world. as part of
                                                                                                    another geographic information systems
    20
    alGorithms faculty : dePartMent OF COMPuter sCienCe at duke university
                                                                                                                                                                                 Geometric computinG


                                                                                                                                                                                 BioloGical computinG




education:                               Professor Herbert edelsbrunner’s research        protein structure is critical. studying the
Ph.d., Graz university of technology,    focuses on developing algorithms to model        human genome and the proteins that
austria, 1982                            proteins as geometric shapes to determine        make up genes will help us understand
M.s., Graz university of technology,     how structure affects their function. in         the biological mechanisms of illness. the
austria, 1980                            a recent research milestone Professor            result will lead to a new era of molecular
                                         edelsbrunner, one of the founders of             medicine, with new and better ways to
research interests:                      computational geometry, discovered how           diagnose and treat diseases precisely.
algorithms, computational geometry,      to express and compute mathematically              in 1996, Professor edelsbrunner and his
computational topology, and              the weighted area and volume and their           wife Ping Fu founded raindrop Geomagic                       herBert edelsBrunner
computational structural biology.        first and second derivatives. they can           inc. the start-up provides software that                          arts and sciences Professor
                                         now be used to estimate the hydrophobic          automatically converts data scanned from                                 of Computer science
current projects:                        contribution to the driving force of a protein   a physical object, such as a cylinder head                          Professor of Mathematics
Genome-wide analysis of root traits;     in motion. although there are thousands          or a molded hearing aid, to production
algebraic topological tools for high-    of researchers who focus on the study of         quality, 3d digital models.
dimensional analysis; Microstates        protein structure, Professor edelsbrunner
to macrodynamics: a new mathematics      is one of a handful of computer scientists       selected publications:
of biology.                              devoted to this field. By leveraging his         + edelsbrunner, H., and Harer, J. “Persistent
                                         training as a computer scientist and               homology — a survey.” in Surveys on
honors and awards:                       mathematician, his research into protein           Discrete and Computational Geometry.
Member of the German academy of          structure gives biologists and chemists a          Twenty Years Later. Contemporary
science, 2008; Honorary doctorate,       different perspective of structure.                Mathematics, Goodman, J. e., Pach, J.,
Graz university of technology, 2006;       in 2000, Professor edelsbrunner received         and Pollack, r., eds. amer. Math. soc.:
Member of the american academy of arts   a $7.2 million grant from the national             Providence, rhode island, 2008.
and sciences, 2005; national science     science Foundation to collaborate with           + Cohen-steiner, d., edelsbrunner, H.,
Foundation Waterman award, 1991.         stanford university, the university of             and Harer, J. “stability of persistence
                                         north Carolina at Chapel Hill, and north           diagrams.” Discrete Comput. Geom.,
                                         Carolina a&t university on research into           37, 2007.
                                         bioinformatics, applying information             + edelsbrunner, H. “surface reconstruction
                                         technology to solve the riddles of protein         by wrapping finite sets in space.” in
                                         structure. there are numerous medical              Discrete and Computational Geometry —
                                         applications for his research. since the           The Goodman-Pollack Festschrift, aronov,
                                         completion of the Human Genome Project to          B., Basu, s., Pach, J., and sharir, M. eds.
                                         map and sequence the human genetic code,           springer-verlag: Berlin, 2003.
                                         collaborators realize that understanding                                                                                                               21
                                                                                                                      dePartMent OF COMPuter sCienCe at duke university :   alGorithms faculty
memory systems and massive data manaGement


BioloGical computinG




                                                        education:                                 Professor thomas LaBean’s research focuses    using dna as a “smart material” to form
                                                        Ph.d., university of Pennsylvania, 1993    on three aspects of dna engineering:          specific pattern structures with nanometer
                                                        B.s., Honors College, Michigan state       self-assembly of dna nanostructures           scale feature resolution, which will be
                                                        university, 1986                           for biomolecular computing, nano-             used to organize other materials, such
                                                                                                   fabrication, and the use of dna as a          as metals and carbon nanotubes. this
                                                        research interests:                        databasing material. since silicon-based      ability to make smaller patterns surpasses
                                                        dna-based computing, self-assembling       microprocessors are close to hitting their    existing capabilities with lithography
                                                        nanostructures, biologically-inspired      limits of speed and miniaturization,          and opens the door to fabrication on a
         thomas h. laBean                               materials, biomolecular structure,         researchers are exploring the use of dna      scale that wasn’t previously possible.
         associate research Professor                   engineering, and evolution.                (deoxyribonucleic acid) as a new material     if dna can be used to organize organic
         of Computer science                                                                       that will result in faster computing by       molecules for electronics, the result
         associate research Professor                   current projects:                          enabling massively parallel computations.     would be devices that are a factor of
         of Chemistry                                   design, construction, and testing of         Professor LaBean models and designs         1,000 to 10,000 times smaller than what’s
         associate research Professor                   novel dna structures for nanofabrication   dna nanostructures using electronic           currently available. applications include
         of Biomedical engineering                      and computation; development of            computers, then synthesizes the designed      the further miniaturization of electronics,
                                                        self-assembling dna structures for use     oligonucleotides in the lab and anneals       implantable medical devices, biosensors
                                                        as electronic interconnects between        them to form nano scale structures which      and molecular therapeutics.
                                                        molecular-scale and micron-scale           perform calculations during the process of
                                                        devices and electrodes; integration        self-assembly. in 2000, he and a group of     selected publications:
                                                        of nanoparticulate matter and carbon       researchers at duke and nyu successfully      + sebba, d., Mock, J.J., smith, d.r., LaBean,
                                                        nanotubes into novel materials via dna     implemented an XOr computation using            t.H., and Lazarides, a.a. “reconfigurable
                                                        self-assembly; development of computer     dna. they’re now modifying the tiling           Core-satellite nanoassemblies as
                                                        architecture strategies enabled by dna-    structure so that it can also perform           Molecularly-driven Plasmonic switches.”
                                                        based nanotechnology.                      addition. individual strands of dna             Nano Letters, 8, 2008.
                                                                                                   encode both the inputs and the outputs        + Park, s.H., Finkelstein, G., and LaBean,
                                                                                                   of the computation. “When it works, we’ll       t.H. “stepwise assembly of dna tile
                                                                                                   have the entire look-up table of addition       Lattices using dsdna Bridges.” J. Am.
                                                                                                   encoded in molecules, which then might be       Chem. Soc., 130, 2008.
                                                                                                   useful as inputs for further computations,”   + LaBean, t.H., and Li, H. “using dna for
                                                                                                   said Professor LaBean.                          Construction of novel Materials.” Nano
                                                                                                     in his research on self-assembly for          Today, 2, 2007.
                                                                                                   nano-fabrication, Professor LaBean is

    22
    alGorithms faculty : dePartMent OF COMPuter sCienCe at duke university
                                                                                                                                                                            Geometric computinG


                                                                                                                                                                           BioloGical computinG




education:                                Professor kamesh Munagala’s research        selected publications:
Ph.d., stanford university, 2003          focuses on designing algorithms for         + Guha, s., Munagala, k., and shi, P.
M.s., stanford university, 2002           problems coming from a wide variety of        “approximation algorithms for restless
B.tech., indian institute of technology   application areas. More specifically, he      Bandit Problems.” Proc. ACM-SIAM Symp.
Bombay, 1998                              is interested in approximation algorithms     on Discrete Algorithms (sOda), 2009.
                                          for nP-Hard combinatorial problems that     + Goel, a., and Munagala, k. “Hybrid
research interests:                       arise in routing and power allocation         keyword search auctions.” Proc. Intl.
design and analysis of algorithms;        in computer networks, or computation          World Wide Web Conference (WWW), 2009.
data mining; Computational biology.       in streaming database systems, among        + Munagala, k., srivastava, u., and                                  kamesh munaGala
                                          other applications. the emphasis is           Widom, J. “Optimizing Continuous                                        assistant Professor
honors and awards:                        on algorithms that run in polynomial          Queries with shared expensive Filters.”                                of Computer science
alfred P. sloan Fellow, 2009;             time, and have provable performance           Proc. ACM Symp. on Principles of Database
nsF Career award, 2008.                   guarantees on the quality of the solution     Systems (POds), 2007.
                                          produced, in addition to being simple
                                          to implement.
                                            Professor Munagala also works on
                                          developing new models of computation
                                          for emerging database systems, especially
                                          those arising in streaming applications
                                          and sensor networks.




                                                                                                                                                                                          23
                                                                                                                 dePartMent OF COMPuter sCienCe at duke university   : alGorithms faculty
BioloGical computinG


learninG and modelinG




                                                        education:                                 Professor John reif ’s interdisciplinary      molecular robotics and has developed
                                                        Ph.d., Harvard university, 1977            research includes randomized and              a number of methods for getting dna
                                                        B.s., tufts university, 1973               parallel algorithms, robotics, and            nanostructures to move in a controllable
                                                                                                   molecular nanoassembly and molecular          and autonomous fashion.
                                                        research interests:                        computation. Currently he’s concentrating
                                                        Molecular assembly, dna computing,         his efforts on molecular nanoassembly         selected publications:
                                                        robot motion planning, parallel            and dna computations, for which he has        + yin, P., Hariadi, r.F., sahu, s., Choi,
                                                        algorithms, randomized algorithms, graph   received a number of grants and contracts       H.M.t., Park, s.H., LaBean, t.H.,
         john h. reif                                   algorithms, algebraic computations, data   from federal agencies including darPa           and reif, J.H. “Programming dna
         a. Hollis edens Professor                      compression, and optical computation.      and nsF.                                        tube Circumferences.” Science,
         of Computer science                                                                          the dna molecule is composed of              321(5890), 2008.
                                                        honors and awards:                         sequences of bases that pair up in            + sahu, s., LaBean, t.H., and reif, J.H.
                                                        american association for the               a predictable manner to form dna                “a dna nanotransport device Powered
                                                        advancement of science Fellow, 2003;       nanostructures. Professor reif and his          by Polymerase.” Nano Letters, 8, 2008.
                                                        aCM Fellow, 1997; ieee Fellow, 1993;       research team are conducting research         + reif, J.H., and LaBean, t.H.
                                                        institute of Combinatorics and its         on the use of dna nanostructures to             “autonomous Programmable
                                                        applications Fellow, 1991.                 form patterned lattices and to execute          Biomolecular devices using self-
                                                                                                   computations. the goal of much of this          assembled dna nanostructures.”
                                                                                                   research is to achieve programmed               Communications of the ACM (CaCM),
                                                                                                   patterning of self-assembling molecular         50(9), 2007.
                                                                                                   structures, which can be used in the future
                                                                                                   to pattern molecular electronic circuits.
                                                                                                   this patterning would be at a scale far
                                                                                                   smaller than possible by traditional
                                                                                                   lithographic techniques currently used
                                                                                                   to fabricate computer chips.
                                                                                                      the first demonstration of the
                                                                                                   programmed patterning of bar-code
                                                                                                   patterns at the molecular scale and the
                                                                                                   first demonstration of the conductivity of
                                                                                                   wires constructed from self-assembled
                                                                                                   dna tubes was done by reif ’s group in
                                                                                                   april of 2003. the group also works on
    24
    alGorithms faculty : dePartMent OF COMPuter sCienCe at duke university
s e c o n d a r y f a c u lt y


terrence furey                               john harer                                    mauro maGGioni
assistant research Professor                 Professor of Mathematics                      assistant Professor of Mathematics
of Biostatistics and Bioinformatics          Professor of Computer science                 assistant Professor of Computer science
assistant research Professor
of Computer science

education:                                   education:                                    education:
Ph.d., university of California,             Ph.d., university of California,              Ph.d., Washington university,
santa Cruz, 2002                             Berkeley, 1979                                st. Louis, 2002
M.s., university of California,              B.s., Haverford College, 1974                 M.s., Washington university,
santa Barbara, 1993                                                                        st. Louis, 2000
B.s., university of California,              Professor John Harer’s primary research       Laurea in Matematica universita’ degli
santa Barbara, 1991                          is in the use of algebraic, geometric         studi di Milano, italy, 1999
                                             and combinatorial techniques to study
Professor terrence Furey’s research          the moduli space of algebraic curves          Professor Mauro Maggioni’s primary
focuses on genome sequence analysis          (riemann surfaces). Current research in       research is in harmonic analysis, spectral
with the aim of uncovering mechanisms        this direction is on the homology of the      graph theory, multiscale analysis,
of biological phenomena. Previously,         moduli space, its level covers, and torelli   stochastic dynamical systems, signal
he played a key informatics role in the      space. He also works on the topology          processing, applications to machine
sequencing of the human genome as            of the moduli space of real algebraic         learning, and Markov decision processes.
a member of the international Human          curves. recently, his research focused on     He is also interested in hyperspectral
Genome sequencing Consortium. His            algebraic geometry, differential geometry,    imaging, in particular in building automatic
work helped ensure the production of a       geometric topology, combinatorics,            classifiers for discriminating normal from
highly accurate and essentially complete     computational geometry, robotics, and         cancerous biopsies, and in the geometry
sequence, especially with regard to gene     computational biology.                        of multiscale dynamical systems, and the
sequences. He has also contributed to the                                                  construction of algorithms for the empirical
sequencing and analysis of the mouse, rat,                                                 construction of approximate equations
and chicken genomes.                                                                       for such systems.




                                                                                                                                                                                           25
                                                                                                                      dePartMent OF COMPuter sCienCe at duke university   : alGorithms faculty
artificial
intelliGence
faculty
                                                                                                                                         alGorithms

                                                                                                                                                      artificial intelliGence

                                                                                                                                                                                scientific computinG

                                                                                                                                                                                                       systems and architecture

                                                                                                                                                                                                                                  education
                                                  research projects                                        Geometric computinG
                           the colored chart enables you to easily identify                                                                                                                                                                        internet systems, networkinG, and security
                                   on which research projects our artificial
                                                                                  memory systems and massive data manaGement
                              intelligence group is currently working. each
                               faculty page displays a colored square that                                                                                                                                                                                               BioloGical computinG


                                       corresponds to the research project.                                      learninG and modelinG




artificial intelligence (ai) develops computational methods that meet or exceed the abilities of the human mind
in areas such as perception, reasoning, learning, and planning. the ai group investigates basic data structures
and algorithms for representing, reasoning with, and learning from visual, auditory, and other modes of input.
ai methods have been deployed in countless areas, ranging from commercial desktop applications to nasa space-
craft guidance systems.
   Our ai research focuses on natural language dialog systems, robotics, computer vision, and statistical methods
for learning and planning. the natural-language group is developing sophisticated dialog models that will
incorporate learning and planning techniques to respond efficiently and accurately to verbal requests made by
humans. Computer-vision research focuses on accurate modeling of scenes based on visual data. Our planning and
learning research investigates algorithms for modeling and controlling physical systems, such as robot motion.

                              primary                                          secondary                                                                          emeritus
                              + vincent Conitzer                               + sayan Mukherjee                                                                  + alan W. Biermann
                                (secondary appointment in economics)             (primary appointment in                                                          + donald W. Loveland
                              + Bruce r. donald                                  statistical sciences)
                                (secondary appointment                         + uwe Ohler
                                in Biochemistry)                                 (primary appointment in
                              + alexander Hartemink                              Biostatistics and Bioinformatics)
                              + ronald Parr
                              + Carlo tomasi
                                                                                                                                                                                                                                                                                                27
                                                                                                       dePartMent OF COMPuter sCienCe at duke university                                                                                      : artificial intelliGence faculty
learninG and modelinG




                                                        education:                                  Professor vincent Conitzer’s research          selected publications:
                                                        Ph.d., Carnegie Mellon university, 2006     focuses on issues in the intersection of       + Guo, M., and Conitzer, v. “Worst-Case
                                                        M.s., Carnegie Mellon university, 2003      computer science (especially artificial          Optimal redistribution of vCG Payments
                                                        a.B., Harvard university, 2001              intelligence) and economics. this includes       in Multi-unit auctions.” accepted to
                                                                                                    the design of new marketplaces and other         Games and Economic Behavior, 2009.
                                                        research interests:                         negotiation protocols that allow humans        + Conitzer, v., and sandholm, t. “new
                                                        intersection of artificial intelligence     and software agents to express their             Complexity results about nash
                                                        and economics.                              preferences naturally and accurately,            equilibria.” Games and Economic
         vincent conitzer                                                                           and that generate good outcomes based            Behavior, 63(2), 2008.
         assistant Professor                            current projects:                           on these preferences. it also includes the     + Conitzer, v., sandholm, t., and Lang, J.
         of Computer science                            science applications international          design of software agents that can act           “When are elections with Few Candidates
         assistant Professor                            Corporation (saiC) project to support       strategically in settings where multiple         Hard to Manipulate?” Journal of the ACM,
         of economics                                   voice dialog research; tools for dialogue   parties all pursue their own interests. this     54(3), 2007.
                                                        research (funded by subcontract from        requires the use of concepts from game
                                                        research triangle institute nsF grant).     theory, as well as operationalizing these
                                                                                                    concepts by finding efficient algorithms
                                                        honors and awards:                          for computing the corresponding
                                                        aaai Outstanding Paper award, 2008;         solutions. Finally, his research includes
                                                        alfred P. sloan Fellow, 2008; Honorable     the study of all settings in computer
                                                        mention for the aCM distinguished thesis    science in which multiple parties will act
                                                        award, 2008; Best Program Committee         in their own self-interest, as well as the
                                                        Member award, aaMas 2006; iBM Ph.d.         design of incentive mechanisms to reach
                                                        Fellow, 2005/2006.                          good outcomes in spite of such behavior.




    28
    artificial intelliGence faculty : dePartMent OF COMPuter sCienCe at duke university
                                                                                                                                                                                 Geometric computinG


                                                                                                                                                                                 BioloGical computinG




education:                                  Geometric algorithms (PGa) arise in           selected publications:
Ph.d., Mit, 1987                            developing and applying information           + Georgiev, i., and donald, B.r. “dead-end
s.M., Mit, 1984                             technology to understand the molecular          elimination with Backbone Flexibility.”
B.a., yale, 1980                            machinery of the cell. donald’s recent          Bioinformatics, 23(13), 2007.
                                            work shows that many PGa techniques           + Gorczynski, M.J., donald, B.r., et. al.
major research interests:                   may be fruitfully applied to the challenges     “allosteric inhibition of the Protein-
Computational biology,                      of computational molecular biology.             Protein interaction Between the
microelectromechanical systems,             PGa research should lead to computer            Leukemia-associated Proteins runX1 and
microrobotics, robotics, geometry,          systems and algorithms that are useful in       CBF-beta.” Chem. Biol., 14(10), 2007.                                    Bruce r. donald
graphics and algorithms.                    structural molecular biology, proteomics,     + donald, B.r., Levey, C.G., McGray,                           William and sue Gross Professor
                                            and rational drug design. Concomitantly,        C.d., Paprotny, i., and rus, d. “an                                     of Computer science
honors and awards:                          a wealth of interesting computational           untethered, electrostatic, Globally-                                  Professor of Biochemistry
John simon Guggenheim Memorial              problems arise in proposed methods              Controllable MeMs Micro-robot.”
Fellow, 2001; nsF Presidential young        for discovering new pharmaceuticals.            Journal of Microelectromechanical
investigator, 1989.                         some recent results from the donald             Systems, 15(1), 2006.
                                            laboratory include: new algorithms for
Bruce donald’s interdisciplinary research   interpreting X-ray crystallography and
includes several fields of computational    nMr (nuclear magnetic resonance)
science and engineering, spanning robotics, data, disease classification using mass
geometry, graphics, and algorithms.         spectrometry of human serum, and protein
Currently he is concentrating his efforts   redesign. His algorithms have recently
on two areas, Computational Biology,        been used, respectively, to reveal the
and Microelectromechanical systems          enzymatic architecture of organisms high
(MeMs) and Microrobotics. For example,      on the CdC bioterrorism watch-list, for
his group recently developed the smallest   probabilistic cancer classification from
controllable untethered mobile microrobot, human peripheral blood, and to redesign
by using MeMs technologies. as a primary    an antibiotic-producing enzyme to bind a
faculty member in Computer science,         novel substrate.
donald also is Professor of Biochemistry in   trained at Mit, donald spent 11 years as
the duke university Medical Center.         a professor in the Cornell Computer science
  some of the most challenging and          department. Before coming to duke, he
influential opportunities for Physical      was the Foley Professor at dartmouth.

                                                                                                                                                                                                29
                                                                                                          dePartMent OF COMPuter sCienCe at duke university   : artificial intelliGence faculty
internet systems, networkinG, and security


BioloGical computinG


learninG and modelinG




                                                        education:                                  Professor Hartemink is involved in             in only limited quantities, and 3) coping
                                                        Ph.d., Mit, 2001                            a number of research efforts built             effectively, from both a computational
                                                        s.M., Mit, 1997                             upon a single foundation, namely, the          and a statistical perspective, with the
                                                        M.Phil., Oxford university, 1996            development and application of principled      high dimensionality of the various kinds of
                                                        B.s. and a.B., duke university, 1994        computational and statistical methods          data that abound in this field.
                                                                                                    to elucidate the architecture and
                                                        major research interests:                   function of complex biological systems.        selected publications:
                                                        Computational biology (especially           His work is providing insight on a broad       + Orlando, d., Lin, C., Bernard, a., Wang,
         alexander j. hartemink                         systems biology, systems neurobiology,      range of difficult problems, including           J., socolar, J., iversen, e., Hartemink, a.,
         associate Professor                            transcriptional regulatory networks,        understanding how eukaryotic cells               and Haase, s. “Global Control of Cell-
         of Computer science                            and functional genomics) and machine        regulate the transcription of their genes        Cycle transcription by Coupled Cdk and
                                                        learning (especially Bayesian and dynamic   (with a special focus on transcriptional         network Oscillators.” Nature, 453, 2008.
                                                        Bayesian networks, graphical models,        regulation during the cell cycle in yeast),    + Gordân, r., narlikar, L., and Hartemink,
                                                        supervised and semi-supervised learning,    how networks in the brain are activated          a. “a Fast, alignment-Free, Conservation-
                                                        and Bayesian statistics).                   when processing information or learning          Based Method for transcription Factor
                                                                                                    new tasks (with a special focus on vocal         Binding site discovery.” in Research
                                                        minor research interests:                   learning in songbirds), and how disease          in Computational Molecular Biology
                                                        Molecular and nano computation, human-      can be diagnosed more accurately and the         2008 (reCOMB08). Lecture Notes in
                                                        computer interaction, cryptography, and     usefulness of different therapies can be         Bioinformatics, vingron, M., and Wong, L.,
                                                        computational economics.                    predicted before they are administered           eds. springer: 4955, 2008.
                                                                                                    (with a special focus on using gene and        + Lüdi, P., dietrich, F., Weidman, J.,
                                                        honors and awards:                          protein expression patterns in cancers).         Bosko, J., Jirtle, r., and Hartemink,
                                                        darPa Computer science study Panel,           Professor Hartemink’s research has             a. “Computational and experimental
                                                        2008; david and Janet vaughn Brooks         three main themes: 1) developing                 identification of novel Human imprinted
                                                        teaching award, 2007; alfred P. sloan       machine learning algorithms that are             Genes.” Genome Research, 17, 2007.
                                                        Fellow, 2005; nsF Career, 2004; Orau        fundamentally motivated by a sound
                                                        ralph e. Powe Junior Faculty award, 2002;   understanding of biological systems
                                                        rhodes scholar, 1994; nsF Graduate          but remain statistically principled and
                                                        research Fellow, 1994; Barry M. Goldwater   computationally efficient, 2) producing
                                                        scholar, 1992; Presidential scholar         proper joint learning algorithms that make
                                                        (White House Commission), 1990.             full use of multiple sources of information,
                                                                                                    including prior knowledge, in contexts
                                                                                                    where data are quite noisy or available
    30
    artificial intelliGence faculty : dePartMent OF COMPuter sCienCe at duke university
                                                                                                                                                                                  learninG and modelinG




education:                                   Professor ronald Parr’s research focuses       selected publications:
Ph.d., university of California,             on complex problems that involve               + Parr, r., Li, L., taylor, G., Painter-
Berkeley, 1998                               uncertainty. His work combines probability       Wakefield, C., and Littman, M.
a.B., (cum laude) Princeton                  and decision theory with machine learning        “an analysis of Linear Models, Linear
university, 1990                             techniques to develop policies for acting in     value-Function approximation, and
                                             dynamic and uncertain environments. He is        Feature selection for reinforcement
research interests:                          interested in applying probabilistic models      Learning.” International Conference on
reasoning under uncertainty, Markov          such as Markov decision Processes and            Machine Learning (iCLM), 2008.
decision processes, reinforcement            Bayesian networks to achieve these ends.       + eliazar, a.i., and Parr, r. “Hierarchical                                  ronald e. parr
learning, Bayesian networks, and robotics.     His interests include using reinforcement      Linear/Constant time sLaM using                                             associate Professor
                                             learning to improve the performance of           Particle Filters for dense Maps.”                                          of Computer science
honors and awards:                           systems over time. Professor Parr and his        Advances in Neural Information Processing
iJCai-Jair Best Paper award, 2007;           coauthors have developed new algorithms          Systems (niPs-19), 2005.
nsF Career, 2006; alfred P. sloan Fellow,    that have learned such tasks as bicycle        + Lagoudakis, M., and Parr, r.
2003; Best student Paper award, uai, 2001    riding and soccer-playing in simulations.        “Least-squares Policy iteration.”
with uri Lerner (student first author).      these algorithms can learn to perform well       Journal of Machine Learning Research
                                             after making a relatively small number           (JMLr), 4, 2003.
                                             of observations, in contrast to earlier
                                             approaches that required a huge number
                                             of trial and error experiences.
                                               Professor Parr is also interested in
                                             robotic exploration and mapping. With
                                             austin eliazar, he developed a new
                                             algorithm called dP-sLaM which can
                                             efficiently and accurately produce maps
                                             of new environments using
                                             a laser rangefinder.




                                                                                                                                                                                                  31
                                                                                                            dePartMent OF COMPuter sCienCe at duke university :   artificial intelliGence faculty
                                                    education:                                Professor Carlo tomasi’s research focuses       led to systems for the automatic detection
                                                    Ph.d., Carnegie Mellon university, 1991   on the analysis of image motion, stereo         of colon cancer in its early stages,
                                                    doctorate, university of Padova,          vision, image retrieval, and medical            pointing the way to the automated and
                                                    italy, 1987                               imaging. He made a widely recognized            unobtrusive screening of large numbers
                                                    M.s., university of Massachusetts         contribution to structure from motion with      of people at a relatively low cost. “it feels
                                                    at amherst, 1984                          his Ph.d. thesis, whose ramifications have      really satisfying when your ideas click
                                                                                              generated a steady flow of publications         from a theoretical standpoint,” says
                                                    research interests:                       from his research group and others. ten         Professor tomasi. “But when they
     carlo tomasi                                   Computer vision, medical imaging,         years later, his theory of factorization        help save people’s lives, the thrill is
     Professor of Computer science                  and applied mathematics.                  still generates Ph.d. theses across the         incomparably more intense!”
                                                                                              world, has spawned two companies
                                                    current projects:                         (Geometrix in sunnyvale and Point Cloud         selected publications:
                                                    image recognition (funded by nsF);        in Minneapolis), is one of the most cited       + Jiang, t., and tomasi, C. “robust
                                                    stereo and image motion analysis          pieces of work in computer vision, and is         shape normalization based on implicit
                                                    (funded by saiC); Medical imaging         constantly used as a starting point for new       representations.” International
                                                    and computer-assisted diagnosis;          ideas by researchers in the field. in image       Conference on Pattern Recognition, 2008.
                                                    Object recognition.                       retrieval, Professor tomasi has contributed     + tomasi, C. “Global stereo in Polynomial
                                                                                              pioneering concepts and techniques for            time.” In Computational Vision in Neural
                                                                                              the flexible description of image similarity,     and Machine Systems, L. Harris and M.
                                                                                              an obviously fundamental notion in this           Jenkin, eds. Cambridge university Press:
                                                                                              field. His most well-known piece of work          Cambridge, 2007.
                                                                                              in this area is what is widely known as         + Birchfield, s. t., natarajan, B., and
                                                                                              the “earth-Mover’s distance” (eMd), a             tomasi, C. “Correspondence as energy-
                                                                                              flexible and perceptually sound measure of        based segmentation.” Image and Vision
                                                                                              similarity between arbitrary distributions,       Computing, 25(8), 2007.
                                                                                              developed together with yossi rubner and
                                                                                              Leonidas Guibas, and described in a book
                                                                                              Professor tomasi coauthored with yossi
                                                                                              rubner. the generality of this concept has
                                                                                              led to its application also to areas outside
                                                                                              image retrieval, and even outside vision.
                                                                                                Professor tomasi’s recent work in
                                                                                              computer-assisted medical diagnosis has
32
artificial intelliGence faculty : dePartMent OF COMPuter sCienCe at duke university
s e c o n d a r y f a c u lt y                                                              e m e r i t u s f a c u lt y

sayan mukherjee                                uwe ohler                                    alan w. Biermann                                       donald w. loveland
assistant Professor of the institute           assistant Professor                          Professor emeritus                                     Professor emeritus
for Genome sciences and Policy                 of Biostatistics and Bioinformatics          of Computer science                                    of Computer science
assistant Professor of Computer science        assistant Professor of Computer science


education:                                     education:                                   education:                                             education:
Ph.d., Mit, 2001                               Ph.d., Friedrich-alexander-                  Ph.d., university of                                   Ph.d., new york university, 1964
M.s., Columbia university, 1995                universitat, 2001                            California, Berkeley, 1968                             s.M., Massachusetts institute of
B.s.e., Princeton university, 1992             B.s., Friedrich-alexander-                   M.s., Ohio state university, 1961                      technology, 1958
                                               universitat, 1996                            B.e.e., Ohio state university, 1961                    a.B., Oberlin College, 1956
Professor sayan Mukherjee’s research
focuses on computational biology and           Professor uwe Ohler’s research               Professor Biermann’s research focused on               a pioneer in the area of automated
machine learning/statistical learning          contributes to the understanding of          developing a class of natural language                 theorem proving, dr. donald Loveland
theory. in computational biology, his          the regulation of gene expression, i.e.      processors that enables a human to                     was interested in the use of computers to
primary interest is in the analysis for gene   how the full repertoire of genes in an       collaborate efficiently with a machine                 discover the truth of proposed theorems.
expression data for prediction of clinical     organism is active only in the right place   through spoken language. His research                  although standard areas of mathematics
outcomes and to build statistical models       at the right time. to this aim, he uses      stressed the importance of fast,                       were of special interest, proving theorems
incorporating pathway information. in          approaches from pattern recognition, in      convenient communication, in which there               in any theory that can be formally defined
machine learning, his focus is on the          particular probabilistic models, to make     are many interactions per minute. the work             was also of great interest. He studied
interface of statistical learning theory and   sense of the vast amount of data coming      involved the development of a model for a              the Model elimination procedure, a proof
Bayesian statistics, specifically exploiting   from molecular biology today. Projects so    human-computer dialogue system which                   procedure for first-order logic, which used
geometric properties of data to improve        far have identified regulatory genes and     processes subdialogues, uses expectation               the implementation techniques of the
the performance of statistical methods.        control regions in genomic dna of several    to error-correct speech recognition                    logic programming language Prolog. the
                                               organisms (worm, fly, human), but he         problems, adjusts its level of initiative to           implementation was impressively fast but
                                               plans to extend this to other types of       maximize efficiency, and employs a user                computed much redundant information.
                                               data such as images or measurements of       model to properly adjust its interactions              He sought to reduce this redundancy and
                                               gene expression levels.                      for its dialogue partner.                              add an automated lemma capability.



                                                                                                                                                                                                  33
                                                                                                                dePartMent OF COMPuter sCienCe at duke university :   artificial intelliGence faculty
scientific
computinG
faculty
                                                                                                                                         alGorithms

                                                                                                                                                      artificial intelliGence

                                                                                                                                                                                scientific computinG

                                                                                                                                                                                                       systems and architecture

                                                                                                                                                                                                                                  education
                                                  research projects                                        Geometric computinG
                            the colored chart enables you to easily identify                                                                                                                                                                    internet systems, networkinG, and security
                                   on which research projects our scientific
                                                                                  memory systems and massive data manaGement
                                Computing group is currently working. each
                                faculty page displays a colored square that                                                                                                                                                                                           BioloGical computinG


                                       corresponds to the research project.                                      learninG and modelinG




in the last several decades, computers have dramatically changed the way scientific and technological problems
are solved. Computers make it possible to simulate thousands of potential ideas and designs for an object without
actually building the object. Computer and microelectronics, 3d design, drug design, weather forecasting, fusion
machines, and medical device development depend on the ability to simulate a system and its related operating
parameters by computer. key to this approach is the use of numerical mathematical techniques such as the
numerical solutions to partial differential equations and the solution of problems in numerical linear algebra.
   the scientific computing group is widely noted for its research in circuit and semiconductor device simulations,
fluid dynamics, nano-system simulations, and simulations of molecules and fluids with applications to turbulence
and chaos. Our techniques for circuit simulation are applied to modeling the electrical field in the human heart,
in order to improve the effectiveness of life-saving biomedical devices. there is particular interest in applying the
latest in scalable parallel processors to achieve the proposed goals.

                               primary                                         secondary                                                                          emeritus
                               + donald J. rose                                + John a. Board                                                                    + thomas Gallie
                                 (secondary appointment                          (primary appointment in                                                          + Merrell L. Patrick
                                 in Mathematics)                                 electrical and Computer engineering)                                             + C. Frank starmer
                               + Xiaobai sun                                   + Craig Henriquez
                                                                                 (primary appointment in
                                                                                 Biomedical engineering)

                                                                                                                                                                                                                                                                                             35
                                                                                                           dePartMent OF COMPuter sCienCe at duke university                                                                                  : scientific computinG faculty
learninG and modelinG




                                                      education:                                  scientific computing is the synthesis          Professor rose has also been involved in
                                                      Ph.d., Harvard university, 1970             of the mathematics, science, software        novel studies in computational biology,
                                                      a.M., Harvard university, 1967              development, and art needed to solve         such as developing computational models
                                                      B.a., university of California,             major technological problems. in             for studying the effect of drugs on heart
                                                      Berkeley, 1966                              the last several decades, computers          rhythm, and the modeling of MeMs devices
                                                                                                  have dramatically changed the way            for application to the problem of real-
                                                      research interests:                         technological problems are solved.           time gene sequencing. He is currently
                                                      numerical solution of nonlinear algebraic   it is now possible to rapidly simulate       working with duke chemists on the design
        donald j. rose                                and differential equations, numerical       thousands of possible ideas and designs      of molecular structures.
        Professor of Computer science                 linear algebra, and scientific computing.   for an object without actually building
        Professor of Mathematics                                                                  the object.                                  selected publications:
                                                      current projects:                             the numerical solution of differential     + ying, W., rose, d.J., and Henriquez, C.s.
                                                      design of Optimized Materials; inverse      equations focuses on methodology for           “Fully implicit time integration methods
                                                      design of new Molecular structures          solving algebraic differential systems and     for modeling the electrical activity
                                                      (Chemistry and Computer science).           systems of partial differential equations.     of the heart.” technical report, duke
                                                                                                  Motivation comes from vLsi device              university, 2007; IEEE. Trans. Biomed.
                                                                                                  and circuit simulation, control theory         Eng., 55(12), 2008.
                                                                                                  (quantum optimal control), computer          + shao, H., sampson, k.J., Pormann,
                                                                                                  simulations of cardiac electrophysiology,      J. B., rose, d.J., and Henriquez, C. s.
                                                                                                  and optimization.                              “a resistor interpretation of General
                                                                                                    Mathematical foundations are provided        anisotropic Cardiac tissue: use of
                                                                                                  by research on numerical linear algebra        triangular meshes.” Mathematical
                                                                                                  and numerical solutions of nonlinear           Biosciences, 130, 2004.
                                                                                                  algebraic and differential equations.        + rose, d.J. “a graph-theoretic study of
                                                                                                    Professor donald rose works on               the numerical solution of sparse positive
                                                                                                  numerical linear algebra involving             definite systems of linear equations.” in
                                                                                                  research into sparse direct methods            Graph Theory And Computing, read, r.,
                                                                                                  for solving linear systems of equations,       ed. academic Press: new york, 1972.
                                                                                                  nested iteration and block iteration
                                                                                                  methods for linear systems, and issues
                                                                                                  related to implementing these methods
                                                                                                  on novel architecture machines including
                                                                                                  parallel machines.
   36
   scientific computinG faculty : dePartMent OF COMPuter sCienCe at duke university
                                                                                                                                                                               learninG and modelinG




education:                                  to solve larger and more complex problems    developed, and the development of a
Ph.d., university of Maryland               arising in computational science and         parallel counterpart is in progress.
at College Park, 1991                       engineering, numerical algorithms are re-      Matrix theory has been found to
M.s., academia sinica,                      examined in multiple performance aspects     be powerful in their work. to simplify
Beijing, China, 1983                        such as real-time efficiency, scalability,   the analysis of block algorithms with
                                            and programmability, as well as numerical    orthogonal transformations, they have
research interests:                         accuracy and stability. Professor Xiaobai    developed a uniform representation,
numerical analysis, matrix theory,          sun’s recent work has been focused           called the basis-kernel representation, for
high-performance scientific computing,      on understanding, characterizing and         general orthogonal matrices. to ease and                                          xiaoBai sun
and parallel computing.                     developing new and fast algorithms           improve the implementation of blocked                                         associate Professor
                                            to solve large matrix computation            algorithms for high-performance, they have                                   of Computer science
current projects:                           problems, removing obstacles in the          developed a unified theory for aggregating
theory and algorithm development for        dogmatic use of conventional/convenient      various transformations used in matrix
large matrix computation problems arising   assumptions and approaches to handle         computations. numerical algorithm
in computational science and engineering.   emerging, non-traditional problems.          development and computing environment
                                            the emerging computational problems          development have mutual impacts on
                                            include, in particular, those arising from   each other. Professor sun is also involved
                                            Professor sun’s collaboration work with      in cooperative research on programming
                                            other researchers on image restoration       support and system support for high-
                                            and identification, telecommunication        performance matrix computations.
                                            network performance analysis, design and
                                            development of integrated sensing and        selected publications:
                                            processing systems.                          + Brady, d.J., Pitsianis, n.P., and sun,
                                              Professor sun’s research has contributed     X. “reference structure tomography.”
                                            to the design of some new block algorithms     Journal of the Optical Society of America,
                                            and parallel algorithms. For instance,         a 21(7), 2004.
                                            she and her colleagues have developed        + Pauca, v.P., rodriguez, a.F., sun, X., and
                                            a framework, called the successive band        trivedi, k.s. “a Methodology towards
                                            reduction (sBr) approach, for efficiently      automatic implementation of n-body
                                            reducing a large matrix to a condensed         algorithms.” Appl. Num. Math., 40, 2002.
                                            form on high-performance computers with      + sun, X., and Pitsianis, n. “a Matrix
                                            memory hierarchy. a software package           version of the Fast Multipole Method.”
                                            for the serial implementation of sBr is        SIAM Review, 43(2), 2001.
                                                                                                                                                                                               37
                                                                                                            dePartMent OF COMPuter sCienCe at duke university   : scientific computinG faculty
s e c o n d a r y f a c u lt y                                                                    e m e r i t u s f a c u lt y


john a. Board                                      craiG henriquez                                thomas m. Gallie
associate Professor of                             Jeffrey n. vinik Professor                     Professor emeritus
electrical and Computer engineering                of Biomedical engineering                      of Computer science
associate Professor of Computer science            Professor of Computer science


education:                                          education:                                    education:
d.Phil., Oxford university, 1986                    Ph.d., duke university, 1988                  Ph.d. rice university, 1954
M.s., duke university, 1982                         B.s.e., duke university, 1981                 M. a. university of texas, 1949
B.s.e., duke university, 1982                                                                     B. a. Harvard university, 1947
                                                    Professor Craig Henriquez is currently
Professor John Board is interested in               working on the application of the             dr. Gallie’s research was mainly in
applying appropriate high performance               bidomain model to diseased heart tissue       numerical analysis and scientific
computing technology to large and                   to investigate how changes in tissue          computing. these interests included
important scientific or industrial computing        structure (both natural and diseased          inverse problems and other “ill-posed”
problems. His group combines tenets                 induced) and changes in ionic current         problems and their applications to
of computer science and computer                    flow influences the nature of conduction      electrocardiology and geophysics.
engineering with physical insight in the            and the onset of arrhythmia. He is              Other publications were on compiler
development of new algorithms and the               also interested in developing realistic       construction, real-time computing,
implementation of these algorithms on               models that will enable investigators         computer networks for education and
HPC platforms. Much of the work involves            to better interpret electrophysiological      research, computer graphics, and
adapting algorithms for implementation on           measurements made in the clinic. For          pure mathematics.
emerging parallel computing platforms.              example, activation maps at the surface
He is interested in loosely coupled                 of the heart are typically constructed
distributed computing systems (i.e.,                based on the detection of specific features
networked clusters) and in tightly coupled          of the surface extra cellular recordings.
parallel platforms such as the silicon
Graphics Power Challenge, and the Cray t3d.




38
scientific computinG faculty : dePartMent OF COMPuter sCienCe at duke university
merrell l. patrick                                                                         c. frank starmer
Professor emeritus                                                                         Professor emeritus
of Computer science                                                                        of Computer science



education:                                   solving large, sparse systems of equations    education:                                         the institute for nonlinear studies, nice
Ph.d., Carnegie Mellon university, 1964      that arise from the generalized eigenvalue    Ph.d., university of north Carolina                France), Leo rosenshtrakh (Cardiology
M.s., university of kentucky, 1956           problem of structural analysis. Particular    at Chapel Hill, 1968                               Center, Moscow), Jorg Weirich (university
B.s., eastern kentucky university, 1955      attention was given to developing             M.s., duke university, 1965                        of Freiburg, Germany), tassos Bountis
                                             methods, first, for vector processing         B.s., duke university, 1963                        (university of Patras, Greece) and kalyana
dr. Patrick’s early research focused on      systems and later for homogeneous multi-                                                         krishnan (iit-Madras, Chennai india).
the classical problem of finding zeroes      processor systems. spin off efforts focused   dr. starmer’s primary research was at the          in addition to his clinical database work,
of polynomials. this work led to the         on communication requirements of the          boundary of Computer science and Biology,          starmer developed the first physical
development of a family of root finding      algorithms being considered and how to        primarily Cardiology. during the early             model of how drugs bind to ion channels,
methods that contained well-known            partition problems so as to balance the       days of the Cs department, he explored             the first physical characterization of
classical methods as special cases. He       communication costs with the computation      database and operating system design.              vulnerability in an excitable system,
then moved to developing parallel root       costs of candidate parallel systems.          With his colleagues in the department              both biological and chemical.
finding methods that could be used           also parallel architectures and parallel      of Medicine (Cardiology) he established
on then emerging vector and parallel         programming languages suitable for            the computational infrastructure for
processing systems. it is well known that    conducting structural analysis were studied   the duke Cardiology data Bank. He was
the solution of eigenvalue problems can be   and modeled with several colleagues.          fascinated with how cells communicate and
cast as finding roots of the corresponding                                                 how cells perform computations, such as
characteristic equations, but that this                                                    edge detection, maintain regular cardiac
approach is not practical numerically.                                                     rhythms, etc. Here, he collaborated with
Because of this and collaborations with                                                    a.O. Grant in the duke Cardiology division
structural engineers, research efforts of                                                  as well as valentin krinsky (Biophysics
dr. Patrick and his students focused on                                                    institute, Pushchino russia, and now at




                                                                                                                                                                                           39
                                                                                                            department of computer science at duke university   : scientific computinG faculty
systems and
architecture
faculty
                                                                                                                                        alGorithms

                                                                                                                                                     artificial intelliGence

                                                                                                                                                                               scientific computinG

                                                                                                                                                                                                      systems and architecture

                                                                                                                                                                                                                                 education
                                                 research projects                                        Geometric computinG
                           the colored chart enables you to easily identify                                                                                                                                                                   internet systems, networkinG, and security
                               on which research projects our systems and
                                                                                 memory systems and massive data manaGement
                             architecture group is currently working. each
                               faculty page displays a colored square that                                                                                                                                                                                          BioloGical computinG


                                      corresponds to the research project.                                      learninG and modelinG




Our highly regarded systems and architecture research group works on high-impact problems related to providing
users with advanced computing and networking support as the internet continues to permeate our daily lives. the
group covers topics ranging from internet-based services to ad hoc wireless networks, from massive data storage
centers to tiny sensor nodes, from the dataflow within hardware components to global replication of databases across
the internet, and from security to energy efficiency. One major area of research investigates techniques for building
large-scale distributed services that dynamically adapt to widely varying network and user access characteristics in
order to deliver peak levels of performance and reliability.
   the group has been at the forefront of the fundamental aspects of high-impact, innovative research into defining
the foundations of wide-area computing, cooperative management of the memory hierarchy, energy-aware
computing systems, high-speed network-based data storage systems, and nano-scale computer architectures.
                              primary                                         secondary
                              + shivnath Babu                                 + Chris dwyer                                                                        + kishor s. trivedi
                              + Jeffrey s. Chase                                (primary appointment in                                                              (primary appointment in
                              + Landon P. Cox                                   electrical and Computer engineering)                                                 electrical and Computer engineering)
                              + Gershon kedem                                 + romit roy Choudhury
                                (secondary appointment in                       (primary appointment in                                                            emeritus
                                electrical and Computer engineering)            electrical and Computer engineering)                                               + Carla schlatter ellis
                              + alvin r. Lebeck                               + daniel J. sorin                                                                    + robert a. Wagner
                                (secondary appointment in                       (primary appointment in
                                electrical and Computer engineering)            electrical and Computer engineering)
                              + Jun yang
                                                                                                                                                                                                                                                                                           41
                              + Xiaowei yang
                                                                                                 dePartMent OF COMPuter sCienCe at duke university :                                                                                    systems and architecture faculty
internet systems, networkinG, and security


memory systems and massive data manaGement


learninG and modelinG




                                                      education:                                  data management and query processing          selected publications:
                                                      Ph.d., stanford university, 2005            in the presence of multiple, continuous,      + duan, s., Babu, s., and Munagala, k.
                                                      B.tech., indian institute of technology     rapid, time-varying data streams. He            “Fa: a system for automating
                                                      Madras, 1999                                is attacking problems ranging from              Failure diagnosis.” Proc. of the IEEE
                                                                                                  algorithms to data systems in order to          International Conference on Data
                                                      research interests:                         provide a comprehensive solution to data        Engineering (iCde), 2009.
                                                      database systems, architectures             stream management.                            + Babu, s., Borisov, n., uttamchandani,
                                                      and algorithms for self-managing              Computer system performance is another        s., routray, r., and singh, a. “diads:
         shivnath BaBu                                database systems, data management           field where Professor Babu’s research           addressing the My-Problem-or-yours
         assistant Professor                          for new application domains,                promises significant improvement.               syndrome with integrated san and
         of Computer science                          autonomic computing.                        the increasing complexity, scale, and           database diagnosis.” Proc. of the
                                                                                                  dynamics of networked computing                 USENIX Conference on File and Storage
                                                      honors and awards:                          systems make it hard for users and system       Technologies (Fast), 2009.
                                                      nsF Career award, 2007; iBM Faculty         administrators to understand and control      + duan, s., and Babu, s. “Processing
                                                      award, 2006 and 2007; Microsoft Graduate    these systems. recent studies indicate          Forecasting Queries.” Proc. of the
                                                      Fellowship, 2000.                           that a significant fraction of user time        International Conference on Very Large
                                                                                                  gets wasted because of unexpected system        Databases (vLdB), 2007.
                                                      Professor shivnath Babu’s research          slowdowns, crashes, and application
                                                      focuses on database management and          errors. Business-critical systems often
                                                      the improvement of computer system          have hundreds of components, including
                                                      performance. in applications such as        applications, databases, servers,
                                                      network monitoring, telecommunications      routers, etc. and performance relies on
                                                      data management, clickstream                thousands of intricate and time-varying
                                                      monitoring, sensor networks, and others,    dependencies and parameters. Professor
                                                      data takes the form of continuous data      Babu’s research takes the information
                                                      streams rather than finite stored data      streaming in from a system’s myriad
                                                      sets. traditional database systems          components and applies data stream
                                                      and data processing algorithms are ill-     management principles in order to improve
                                                      equipped to handle complex continuous       the understanding, control, and reliability
                                                      queries over data streams. in this          of these complex systems.
                                                      environment, many aspects of data
                                                      management and processing need to be
                                                      reconsidered. Professor Babu investigates
    42
    systems and architecture faculty : dePartMent OF COMPuter sCienCe at duke university
                                                                                                                                                               internet systems, networkinG, and security


                                                                                                                                                             memory systems and massive data manaGement




education:                                 the amount of stored data in the world         Professor Chase manages the internet
Ph.d., university of Washington,           is growing faster than the internet,         systems and storage Group lab, which
seattle, 1995                              making the organization, accessibility,      maintains a 300-node compute cluster
B.a., dartmouth College, 1985              and management of this amount of data        (the devil Cluster) for a wide range of
                                           challenging. Much of this data resides       research projects.
research interests:                        in data centers for access through high-
utility computing, network storage         speed networks. Professor Jeffrey Chase’s    selected publications:
and network i/O, distributed systems,      research explores automated management       + yumerefendi, a., and Chase, J. “strong
operating systems, and large-scale         of data centers and their services as          accountability for network storage.”                                       jeffrey s. chase
network services.                          “utilities” whose resources are delivered      ACM Transactions on Storage (tOs), 3(3),                         Professor of Computer science
                                           to match demand, and sold much like            2007.
current projects:                          electricity is today. He develops systems    + ramakrishnan, L., Grit, L., iamnitchi, a.,
secure Highly available resource           for dynamic resource management,               irwin, d., yumerefendi, a., and Chase, J.
Peering (sHarP) for dependable internet-   scalable storage, and high-speed               “toward a doctrine of Containment: Grid
scale resource management; Cluster-        networking to realize this vision.             Hosting with adaptive resource Control.”
on-demand (COd) cluster automation;          Current research in utility computing        Supercomputing, 2006.
internet emulation (Modelnet);             explores adaptive resource provisioning      + irwin, d., Chase, J., Grit, L.,
request routing for network services       in shared data centers, such as managed        yumerefendi, a., Becker, d., and
(anypoint); active storage and network     Web-hosting facilities. Professor Chase        yocum, k. “sharing networked
storage utilities.                         and his students are building a system         resources with Brokered Leases.”
                                           that continuously monitors and measures        USENIX Technical Conference, 2006.
honors and awards:                         incoming request traffic and server
iBM Faculty award, 2003 and 2004; iBM      performance, and dynamically adjusts
Faculty Partner, 2002; nsF Career award,   the level of server resources to match the
1996; intel Foundation Fellowship, 1993;   load. the system uses advanced network
digital equipment Corporation GeeP         switches that direct incoming request
Fellowship, 1987-1989.                     traffic to selected servers, enabling the
                                           system to allocate server and storage
                                           resources efficiently. the group is also
                                           developing new approaches and software
                                           prototypes to extend dynamic resource
                                           management to internet-scale systems.

                                                                                                                                                                                                    43
                                                                                                     dePartMent OF COMPuter sCienCe at duke university :   systems and architecture faculty
internet systems, networkinG, and security


memory systems and massive data manaGement




                                                       education:                                distributed computing systems were initially   selected publications:
                                                       Ph.d. university of Michigan, 2005        the playground of a well-behaved and well-     + Gaonkar, s., Li, J., roy Choudhury, r.,
                                                       M.s. university of Michigan, 2001         trained few, but the internet and pervasive      Cox, L.P., and schmidt, a. “Micro-Blog:
                                                       B.s. duke university, 1999                wireless technology have extended the            sharing and Querying Content through
                                                                                                 reach of distributed computing to a global       Mobile Phones and social Participation.”
                                                       research interests:                       population of users with a wide range            MobiSys 2008, 2008.
                                                       Cooperative distributed systems, mobile   of goals and levels of sophistication.         + Cox, L.P., dalton, a., and Marupadi, v.
                                                       computing, and operating systems, with    Professor Cox’s research explores the            “smokescreen: Flexible Privacy
          landon cox                                   a focus on privacy and incentives.        challenges and opportunities of these large      Controls for Presence-sharing.”
          assistant Professor                                                                    distributed systems. He is particularly          MobiSys 2007, 2007.
          of Computer science                          current projects:                         interested in two challenges in this domain:   + yumerefendi, a., Mickle, B., and Cox,
                                                       access-control misconfigurations,         designing incentives in cooperative              L.P. “tightLip: keeping applications from
                                                       privacy, and incentives in                distributed systems so that the local goals      spilling the Beans.” NSDI 2007, 2007.
                                                       distributed systems.                      of individual users are aligned with the
                                                                                                 global goals of the system architect as well
                                                       honors and awards:                        as designing software to help users protect
                                                       nsF Career award, 2007.                   themselves from configuration errors that
                                                                                                 can expose their most sensitive data.
                                                                                                   to explore the first challenge, Professor
                                                                                                 Cox and his students are building
                                                                                                 cooperative mobile systems that try to
                                                                                                 balance users’ needs to conserve battery
                                                                                                 power and protect their location privacy
                                                                                                 with the goals of mobile services.
                                                                                                   Professor Cox and his students are also
                                                                                                 building a new operating system called
                                                                                                 tightLip, which allows organizations
                                                                                                 and users to better manage their shared
                                                                                                 spaces by helping them define what data
                                                                                                 is important and who is trusted, rather
                                                                                                 than requiring an understanding of the
                                                                                                 complex dynamics of how data flows among
                                                                                                 software components.
     44
     systems and architecture faculty : dePartMent OF COMPuter sCienCe at duke university
                                                                                                                                                           internet systems, networkinG, and security


                                                                                                                                                         memory systems and massive data manaGement




education:                              Professor Gershon kedem’s research           system to grant or deny user access. Many
Ph.d., university of Wisconsin,         focuses on computer architecture and         computer break-ins are accomplished
Madison, 1978                           network security. since accessing data is    by circumventing the initial protection
B.s., Hebrew university, 1972           the primary bottleneck in computations,      mechanism that validates users, allowing
                                        Professor kedem is developing a high         the user complete access to the entire
research interests:                     performance memory system that uses          system since all protection is turned
Computer-aided design (Cad) of          a hybrid cache of dynamic raM and            off. By using cryptography to implement
integrated circuits, vLsi- oriented     static raM. unlike a regular cache, the      access rights, each person must have
architectures, and computer security.   combination of the two enables the cache     the key to the file in order to see the file                                  Gershon kedem
                                        to benefit from the density and storage      in clear text. “in some sense, you get                                         associate Professor
                                        capability of the d-raM as well as the       authenticated every time you try to access                                    of Computer science
                                        performance of the s-raM. the result is      a file rather than being authenticated                                    associate Professor
                                        a much larger cache that dramatically        once and then getting to roam freely,”                of electrical and Computer engineering
                                        improves performance by allowing for         kedem says.
                                        more aggressive pre-fetching.                  Professor kedem’s focus on security
                                          as information grows and continues         extends to wireless networks as well. akin
                                        to be stored on computer systems for         to a virtual private network connected
                                        collaboration, security has become the       between a laptop and the base station,
                                        cornerstone of every system. Professor       he is creating a network that will rely on a
                                        kedem is exploring potential security        widely deployed communication protocol
                                        weaknesses that are vulnerable to attack     instead of weak protection from the
                                        and penetration in computer systems.         hardware wireless manufacturer.
                                        He and his team are implementing an
                                        adaptable security policy whereby the        selected publications:
                                        computer monitors the activities in the      + Gehani, a., and kedem, G. “rheostat:
                                        system to detect potential intrusions.         real-time risk Management.” RAID 2004,
                                        When the system detects an attack, it          2004.
                                        modifies existing security policies and      + Gehani, a., and kedem, G. “real-time
                                        reduces capabilities or restricts access       access Control reconfiguration.”
                                        to certain critical areas in the system to     International Infrastructure Survivability
                                        prevent and contain the attack.                Workshop (iisW’04), 2004.
                                          Professor kedem also is building a         + Xu, C., kedem, G., and Gong, F.a. “new
                                        secure file access system that relies on       Procedure for Cryptographic Protocol
                                        cryptography instead of the operating          analysis.” SAM2, 2002.                                                                                  45
                                                                                                 dePartMent OF COMPuter sCienCe at duke university   : systems and architecture faculty
memory systems and massive data manaGement


BioloGical computinG




                                                      education:                                  Professor alvin Lebeck’s current              selected publications:
                                                      Ph.d., university of Wisconsin,             research focuses on three broad aspects       + Pistol, C., dwyer, C., and Lebeck, a.r.
                                                      Madison, 1995                               of computer architecture: nanoscale             “architectural implications of nanoscale
                                                      M.s., university of Wisconsin,              computer architectures, energy                  integrated sensing and Computing.”
                                                      Madison, 1991                               efficient computing, and multiprocessor         Proceedings of the 14th International
                                                      B.s., university of Wisconsin,              performance metrics.                            Conference on Architectural Support for
                                                      Madison, 1989                                 Professor Lebeck is exploring the             Programming Languages and Operating
                                                                                                  design of new computing systems that            Systems (asPLOs ‘09), 2009.
         alvin r. leBeck                              research interests:                         address the challenges and exploit the        + Pistol, C., dwyer, C., and Lebeck, a.r.
         Professor of Computer science                Computer architecture, nano-scale           opportunities of nano-scale electronics.        “nanoscale Optical Computing using
         Professor of electrical and                  systems, memory systems, energy efficient   He is leading a group that seeks to utilize     resonance energy transfer Logic.” IEEE
         Computer engineering                         computing, and multiprocessors.             dna-based nanostructures as a scaffold          Micro, 2008.
                                                                                                  on which to place nanoelectronic devices,     + dwyer, C., and Lebeck, a.r. Introduction
                                                      current projects:                           such as carbon nanotubes.                       to DNA Self-Assembled Computer Design.
                                                      architectures for emerging                    the current trend toward multithreaded        artech House Publishers: 2008.
                                                      nanotechnologies, throughput                processors or multiple processors on a chip
                                                      oriented multiprocessors.                   places new demands on programmers to
                                                                                                  understand application behavior. Professor
                                                      honors and awards:                          Lebeck is working with Professor daniel
                                                      Best Paper award, aCM/ieee international    sorin on new hardware-level performance
                                                      symposium on Microarchitecture, 1998;       metrics that provide significantly more
                                                      nsF Career award, 1997.                     insight for understanding the complex
                                                                                                  behaviors of multithreaded applications,
                                                                                                  such as web servers or databases.




    46
    systems and architecture faculty : dePartMent OF COMPuter sCienCe at duke university
                                                                                                                                                                   internet systems, networkinG, and security


                                                                                                                                                                 memory systems and massive data manaGement




education:                                 Professor Jun yang’s research focuses              even relevant data accumulates over
Ph.d., stanford university, 2001           on managing large databases and                  time, making it difficult to answer
B.a., university of California,            information systems. even with infinite          questions quickly. Professor yang is
Berkeley, 1995                             storage capacity, it’s a significant             looking into techniques that may help
                                           challenge to manage, analyze and mine            alleviate that problem: One focuses on
research interests:                        the vast amounts of data—from satellite          returning the most relevant results first,
database and information                   feeds to dna information to Web content—         leaving a slew of correct but uninteresting
processing systems.                        that are continually being received.             results for later, and the second is to offer
                                           Professor yang is exploring new ways to          an approximate answer quickly, and then                                                    jun yanG
honors and awards:                         process data as it arrives so it can be easily   refine it continuously. Both techniques                                         associate Professor
iBM Faculty award, 2006; nsF Career        and efficiently accessed later for queries.      give users a rough picture of the answer                                       of Computer science
award, 2003; u.C. Berkeley Computer          “simply archiving all data is not              along the way, enabling them to make                                 director of Graduate studies
science division Highest achievement       enough,” Professor yang said. “you have          better decisions about whether more
awards, 1995; u.C. Berkeley Chancellor’s   to make choices about what data is more          searches are needed.
scholar, 1993-1995.                        important, and if the problem gets out
                                           of hand, what are the tricks to quickly          selected publications:
                                           generating answers, perhaps not the most         + Chandramouli, B., yang, J., agarwal,
                                           complete or accurate answer, but some              P.k., yu, a., and Zheng, y. “Prosem:
                                           answer that’s ‘good enough.’”                      scalable Wide-area Publish/subscribe.”
                                             One approach Professor yang has taken            Proceedings of the 2008 ACM SIGMOD
                                           to manage constantly growing data is to            International Conference on Management
                                           process queries incrementally. Queries are         of Data (siGMOd ‘08), 2008.
                                           computed in advance over existing data           + silberstein, a., Puggioni, G., Gelfand, a.,
                                           and their results are kept in a database.          Munagala, k., and yang, J. “suppression
                                           as more data arrives, he refines the               and Failures in sensor networks: a
                                           query results incrementally, instead of            Bayesian approach.” Proceedings of the
                                           computing them again from scratch. the             33rd International Conference on Very
                                           problem is determining what additional             Large Data Bases (vLdB ‘07), 2007.
                                           data to keep around in order to process          + Chandramouli, B., Bond, C.n., Babu,
                                           future updates and queries. Professor              s., and yang, J. “Query suspend and
                                           yang is interested in how temporal                 resume.” Proceedings of the 2007 ACM
                                           database techniques, such as filtering             SIGMOD International Conference on
                                           and expiring of historical data, can be            Management of Data (siGMOd ‘07), 2007.
                                           extended and applied in this setting.                                                                                                                       47
                                                                                                         dePartMent OF COMPuter sCienCe at duke university   : systems and architecture faculty
internet systems, networkinG, and security


memory systems and massive data manaGement




                                                       education:                               today, the internet is a sprawling           information in online interactions through
                                                       Ph.d., Massachusetts institute of        network of networks with public, private     an examination of social networks.
                                                       technology, 2004                         and commercial interests, but it was
                                                       B.e., tsinghua university, 1996          first designed as a tool for academics,      selected publications:
                                                                                                a virtual space to share data. as the        + Liu, X., yang, X., and Lu, y. “to Filter or
                                                       research interests:                      internet developed into a mainstream           to authorize: network-Layer dos defense
                                                       networks and distributed systems,        entity, vulnerabilities in security and        against Multimillion-node Botnets.” ACM
                                                       with an emphasis on protocol,            infrastructure surfaced. Professor yang’s      SIGCOMM, 2008
          xiaowei yanG                                 architecture design and security.        research is oriented toward one goal—        + Liu, X., Li, a., yang, X., and Wetherall, d.
          assistant Professor                                                                   designing the internet of tomorrow, one        “Passport: secure and adoptable source
          of Computer science                          current projects:                        more robust to failures, more resilient to     authentication.” USENIX/ACM NSDI, 2008.
                                                       resistant network architecture design;   attacks, and more trustworthy for users.     + yang, X., Clark, d., and Berger, a.W.
                                                       robust routing design; identity and        in the first of three current projects,      “nira: a new inter-domain routing
                                                       trust in system networks.                Professor yang is designing networks to        architecture.” IEEE/ACM Transactions on
                                                                                                withstand botnet attacks. these “zombie        Network (ton), 15(4), 2007.
                                                                                                machines” compromise computers and
                                                                                                congest networks by co-opting processes
                                                                                                to forward transmissions like spam and
                                                                                                viruses. Professor yang and her students
                                                                                                have already built a prototype network
                                                                                                design capable of effectively blocking
                                                                                                most botnet attacks.
                                                                                                  in a second area of research, Professor
                                                                                                yang and students seek to improve
                                                                                                the performance of internet routers,
                                                                                                which can be plagued by interruptions
                                                                                                in performance and temporary losses of
                                                                                                information. the goal is to build a system
                                                                                                with uninterrupted internet access even
                                                                                                during routing changes.
                                                                                                  Finally, Professor yang and her students
                                                                                                are exploring the trustworthiness of
                                                                                                various internet entities. to do so,
     48                                                                                         they are studying the use of identity
     systems and architecture faculty : dePartMent OF COMPuter sCienCe at duke university
s e c o n d a r y f a c u lt y


chris dwyer                                   romit roy choudhury                         daniel j. sorin                                     kishor s. trivedi
assistant Professor of                        assistant Professor of                      assistant Professor of                              Hudson Professor of
electrical and Computer engineering           electrical and Computer engineering         electrical and Computer engineering                 electrical and Computer engineering
assistant Professor of Computer science       assistant Professor of Computer science     assistant Professor of Computer science             Professor of Computer science
                                                                                                                                              director, Center for advanced Computing
                                                                                                                                              and Communication
education:                                    education:                                  education:                                          education:
Ph.d., university of north Carolina           Ph.d., university of illinois at            Ph.d., university of Wisconsin,                     Ph.d., university of illinois at
at Chapel Hill, 2003                          urbana-Champaign, 2006                      Madison, 2002                                       urbana-Champaign, 1974
M.s., university of north Carolina            M.s., university of illinois at             M.s., university of Wisconsin,                      M.s., university of illinois at
at Chapel Hill, 2000                          urbana-Champaign, 2003                      Madison, , 1998                                     urbana-Champaign, 1972
B.s., Pennsylvania state university, 1998     B.tech., Haldia institute of technology,    B.s.e., duke university, 1996                       B.tech., indian institute of technology,
                                              india, 2000                                                                                     Bombay, 1968
Professor Chris dwyer is currently working                                                Professor daniel sorin’s research focuses
on the design and fabrication of nanoscale    Professor romit roy Choudhury is            on the design and evaluation of shared              Professor kishor trivedi’s research focuses
self-assembling computer systems and          currently on wireless networks, including   memory multiprocessors. these systems               on reliability and performance assessment
architectures. He is also interested in dna   supporting multiple users, managing         run the commercial workloads, such as               of computer and communication systems.
self-assembly, nanoscale circuit design,      mobility of users, providing reliable       databases and web servers, upon which               recent research accomplishments
and design tool and simulator support for     communication in the face of wireless       society increasingly relies. as such, their         include three areas of activity: advances
emerging device technologies.                 channel fluctuations, routing, security,    performance, availability, and design               in modeling techniques; performance,
                                              privacy, among other challenges. He has     complexity are all important issues that            reliability and dependability modeling
                                              worked on systems that exploit smart        Professor sorin addresses in his research.          of applications; and development and
                                              antennas to improve wireless network                                                            dissemination of modeling tools.
                                              performance, location management,
                                              and on sensor network problems.




                                                                                                                                                                                         49
                                                                                                      dePartMent OF COMPuter sCienCe at duke university   : systems and architecture faculty
e m e r i t u s f a c u lt y


carla schlatter ellis                             roBert a. waGner
Professor emeritus of                             Professor emeritus of
Computer science                                  Computer science



education:                                         education:
Ph.d., university of Washington, 1979              Ph.d., Carnegie Mellon university, 1969
M.s., university of Washington, 1977               B.s., Massachusetts institute of
B.s., university of toledo, 1972                   technology, 1962

While at duke, dr. ellis’ research was on          dr. Wagner’s research interests included
operating systems for mobile, parallel,            experimental vLsi architectures,
and distributed systems, with particular           application of dynamic programming to
focus on energy management of computing            algorithms and systems design, design of
devices. she spearheaded the Milly Watt            optimal software and hardware systems,
project to explore how the operating               and time cost trade-offs in abstract
system and architecture of handheld                parallel computer models. in particular,
mobile devices could be redesigned to              his research focused on changing
manage power efficiently. With the help            computer architecture and optimization
of their students, dr. ellis and Professor         of compilers to speed computations.
alvin Lebeck built and refined eCOsystem           Professor Wagner explored methods to
(energy-Centric Operating system), a               improve memory reference latency and
Linux-based operating system that views            processor ipC.
energy as a first-class resource and
manages it explicitly.




50
systems and architecture faculty : dePartMent OF COMPuter sCienCe at duke university
21
education
faculty
                                                                                                                                            alGorithms

                                                                                                                                                         artificial intelliGence

                                                                                                                                                                                   scientific computinG

                                                                                                                                                                                                          systems and architecture

                                                                                                                                                                                                                                     education
                                                research projects                                       Geometric computinG
                                 the colored chart enables you to easily                                                                                                                                                                         internet systems, networkinG, and security
                                  identify on which research projects our
                                                                               memory systems and massive data manaGement
                                   education group is currently working.
                             each faculty page displays a colored square                                                                                                                                                                                               BioloGical computinG


                               that corresponds to the research project.                                      learninG and modelinG




the mission of the education group is to develop state-of-the-art education tools and practices and to integrate
our research and education goals. the group develops traditional and Web-based materials for undergraduate
courses in computer science, including formal languages and automata, and spearheads a department-wide
initiative focused on rapidly integrating departmental research into undergraduate courses. Many of these activities
are funded by grants from industry or government agencies.


                             primary                                        secondary                                                                                emeritus
                             + Owen astrachan                               + timothy Lenoir                                                                         + dietolf ramm
                             + robert C. duvall                               (primary appointment as
                             + Jeffrey r.n. Forbes                            kimberly Jenkins Chair for
                             + richard a. Lucic                               new technologies and society)
                             + susan H. rodger




                                                                                                                                                                                                                                                                                              53
                                                                                                                            dePartMent OF COMPuter sCienCe at duke university                                                                                 : education faculty
                                                    education:                                  as Professor of the Practice, dr. Owen        used independently, but still relates to
                                                    Ph.d., duke university, 1992                astrachan focuses on developing               other modules and concepts.
                                                    M.s., duke university, 1989                 techniques to simplify and communicate          “Packaging is one key in getting others
                                                    M.a.t., duke university, 1979               complex, cutting-edge computer science        to use your research, your software
                                                    a.B., dartmouth College, 1978               research—without sacrificing technical        artifacts, and your course materials. it’s
                                                                                                rigor and accuracy—to students and            critical that researchers and educators
                                                    research interests:                         industry partners. He also develops           learn to package their work effectively,”
                                                    Object-oriented design, software            techniques and technologies for learning      says Professor astrachan.
     owen astrachan                                 architecture, computer science              and mastering programming, particularly
     Professor of the Practice                      education, apprentice learning,             object-oriented programming and agile         selected publications:
     of Computer science                            and automated reasoning.                    software methodologies. at the same time      + alt, C., astrachan, O., Forbes, J., Lucic,
     Co-director of undergraduate studies                                                       that he works on methods for teaching           r., and rodger, s. “social networks
                                                    honors and awards:                          programming, Professor astrachan is             Generate interest in Computer science.”
                                                    nsF Cise distinguished education Fellow,    working on developing non-programming           SIGCSE Technical Symposium on Computer
                                                    2007; richard k. Lublin teaching award,     approaches to computer science,                 Science Education, 2006.
                                                    2002; Outstanding instructor in Computer    especially approaches that emphasize          + astrachan, O. “non-Competitive
                                                    science (university of British Columbia),   contributions of computer science to other      Programming Contests as the Basis for
                                                    1998; nsF Career award, 1997; robert Cox    areas of research.                              Just-in-time teaching.” Proceedings
                                                    teaching award, 1995.                         Professor astrachan develops course           of the Frontiers in Education (FIE)
                                                                                                and curricular materials to deliver             Conference, 2004.
                                                                                                the best instruction and instructional        + astrachan, O. “Bubble sort: an
                                                                                                materials to students at duke, with             archaeological algorithmic analysis.”
                                                                                                the aim of exporting the materials to a         SIGCSE Technical Symposium on Computer
                                                                                                wider audience. this requires developing        Science Education, 2003.
                                                                                                modules that are general enough to be
                                                                                                included and used in a variety of curricula
                                                                                                and settings, while ensuring that the
                                                                                                materials are engaging and relevant in
                                                                                                helping to understand specific concepts.
                                                                                                Challenges include developing cohesive
                                                                                                modules that are not tightly coupled to
                                                                                                each other so that each module can be

54
education faculty : dePartMent OF COMPuter sCienCe at duke university
education:                             robert duvall’s primary duties include        sticky, rubbery substance called icky-poo.
M.sc., Brown university, 1997          teaching and developing techniques and        toys serve as extremely good props and
B.a., Brown university, 1993           tools that make it easier for students to     metaphors, giving students a physical
                                       learn about abstract computer science         image and memorable experience in an
research interests:                    concepts. duvall teaches a variety of         effort to better understand and remember
Object-oriented programming,           undergraduate programming courses,            computer science concepts.
design patterns, graphics, animation   including software design, computer
and visualization, artificial life,    graphics, web programming, data               selected publications:
and programming languages.             structures, and classes for non-majors        + Pollard, s., and duvall, r. “everything                               roBert c. duvall
                                       and majors with no previous programming         i needed to know about teaching i                                                       Lecturer
                                       experience. “i like making computer             Learned in kindergarten: Bringing
                                       science exciting and interesting, not geeky     elementary education techniques
                                       and nerdy,” he said.                            to undergraduate Computer science
                                         to that end, duvall believes in               Classes.” SIGCSE Technical Symposium on
                                       bringing research topics and real-world         Computer Science Education, 2006.
                                       applications into his courses as examples     + astrachan, O., duvall, r., Forbes, J.,
                                       and assignments. this is a challenge            and rodger, s. “active Learning in
                                       because these topics must be presented          small to Large Courses.” IEEE Frontiers in
                                       such that students can understand and           Education Conference, 2002.
                                       appreciate them without losing their          + astrachan, O., duvall, r., and
                                       significance or accuracy. For example,          Wallingford, e. “Bringing extreme
                                       duvall’s introduction to programming            Programming to the Classroom.”
                                       for non-majors course is heavily based          XPUniverse, 2001.
                                       on computer graphics. assignments to
                                       develop games such as space invaders,
                                       asteroids, and tetris add fun to the class,
                                       but also challenge students intellectually
                                       to solve interesting problems.
                                         duvall and his colleagues will utilize
                                       just about anything in order to make
                                       abstract concepts concrete, from flash
                                       cards, candy, and plastic building blocks
                                       originally designed for toddlers, to a
                                                                                                                                                                                          55
                                                                                                                   dePartMent OF COMPuter sCienCe at duke university   : education faculty
learninG and modelinG




                                                        education:                                 as assistant Professor of the Practice,       Forbes. “and it will motivate them to think
                                                        Ph.d., university of                       Jeffrey Forbes develops innovative            logically about their program and discern
                                                        California-Berkeley, 2002                  computer science teaching techniques          its properties.”
                                                        B.s., stanford university, 1993            that are effective and accessible. He           Professor Forbes is also exploring how
                                                                                                   focuses on developing new methods for         theoretical aspects of computer science
                                                        research interests:                        enabling active learning and engaging         such as logic, probability, and other tools in
                                                        Computer science education,                students in classroom situations which        discrete mathematics can be incorporated
                                                        intelligent agents, and network science.   compel them to read, listen, and              into the introductory curriculum.
         jeffrey r.n. forBes                                                                       think deeply and investigates their
         assistant Professor of the                                                                effectiveness. in motivating novice           selected publications:
         Practice of Computer science                                                              students, Forbes wants to communicate         + alt, C., astrachan, O., Forbes, J., Lucic,
                                                                                                   the excitement of computer science              r., and rodger, s. “social networks
                                                                                                   methodology and research.                       Generate interest in Computer science.”
                                                                                                     Professor Jeffrey Forbes’ research            Proceedings of the 36th Technical
                                                                                                   centers on applying artificial intelligence     Symposium on Computer Science
                                                                                                   and machine learning techniques to              Education, 2006.
                                                                                                   real-world situations. One particular         + Bailey, t., and Forbes, J. “Just-in-time
                                                                                                   area of interest is controlling mobile          teaching for Cs0.” Proceedings of the 36th
                                                                                                   robots. Forbes integrates robots into           Technical Symposium on Computer Science
                                                                                                   the undergraduate program to help               Education, 2005.
                                                                                                   students understand computer science          + Bailey, t., and Forbes, J. “Computers and
                                                                                                   concepts and processes. With a team of          society in Cs0: an interactive approach.”
                                                                                                   undergraduate researchers, Forbes is            Proceedings of the 34th ASEE/IEEE
                                                                                                   developing algorithms for robot soccer.         Frontiers in Education Conference, 2004.
                                                                                                   in partnership with neighborhood schools,
                                                                                                   the team is studying the effectiveness of
                                                                                                   using robots as a tool for teaching and
                                                                                                   motivating interest in math, science, and
                                                                                                   computers. “asking students to program
                                                                                                   a robot to perform a certain function will
                                                                                                   help students think about the debugging
                                                                                                   process in a new way,” says Professor

    56
    education faculty : dePartMent OF COMPuter sCienCe at duke university
education:                                 as director of external relations,            and top-notch graduates. He also serves
M.s., stanford university, 1971            richard Lucic manages and facilitates         as Faculty Curriculum director of the
eng. of Metallurgy, Colorado school        relationships with industrial partners,       information sciences and information
of Mines, 1966                             academic institutions, government             studies interdisciplinary program at duke.
                                           agencies and private foundations.               in addition, Lucic teaches classes that
research interests:                        these symbiotic partnerships connect          meld technology with business: Principles
tools for computer science education,      graduating students with leading              of research Management, Principles of
management of research, industrial         companies, acquire research funding,          effective e-Commerce, and Web-based
relations, and technology management       and match faculty expertise with industry     Multimedia Communications.                                              richard a. lucic
and transfer.                              interest for collaborative research. these                                                                      associate Professor of the
                                           relationships foster communication            selected publications:                                         Practice of Computer science
current projects:                          between researchers and industry, and         + alt, C., astrachan, O., Forbes, J., Lucic,                                      associate Chair
Faculty Curriculum director, information   ensure that the department’s research           r., and rodger, s. “social networks               Co-director of undergraduate studies
science and information studies Program;   and curriculum are relevant and timely.         Generate interest in Computer science.”
industry and alumni relations.             Members of the industrial Partners              SIGCSE Technical Symposium on Computer
                                           Program include tech heavyweights Bell          Science Education, 2006.
                                           Labs, Cisco systems, Compaq, Hewlett-         + Lucic, r.a., and Hurka-Owen, k. “tools
                                           Packard, iBM, sun Microsystems, intel           for academic success.” 10th Teaching
                                           and Microsoft.                                  Academic Survival Skills Conference, 1999.
                                             Lucic, whose research of technology         + Lucic, r.a. “Cooperative research and
                                           transfer is used to move technology             technology transfer.” Proc. Technology
                                           effectively from the lab to the users,          Transfer Between Research Institutes and
                                           also manages intellectual property (iP)         Industry, Satellite Symposium of 10th
                                           for the department. to speed the time           Nordic-Baltic Conference on Biomedical
                                           consuming effort of iP management,              Engineering, 1996.
                                           Lucic works with faculty to package their
                                           research for patent submission, as well as
                                           for licensing opportunities with industry
                                           partners. these external interactions are
                                           critical in generating the necessary funds
                                           for duke’s computer science department
                                           to compete with other top-tier universities
                                           as a department with innovative research
                                                                                                                                                                                             57
                                                                                                                      dePartMent OF COMPuter sCienCe at duke university   : education faculty
                                                    education:                                    as Professor of the Practice, Professor        no experience to learn about computer
                                                    Ph.d., Purdue university, 1989                susan rodger’s research focuses on             science. the program involves problem
                                                    B.s., north Carolina state university, 1983   developing tools that help students            solving in small groups and learning to
                                                                                                  visualize and interact with theoretical        program by creating virtual worlds first,
                                                    research interests:                           computer science concepts. For example,        then Java. it is part of a collaborative
                                                    Computer science education, interactive       rodger has helped create JFLaP, a package      effort with seven other institutions.
                                                    and visual tools for teaching, algorithm      of graphical tools that can be used as
                                                    animation, design and analysis of             an aid in learning the basic concepts of       selected publications:
     susan h. rodGer                                algorithms, data structures, geometric        formal languages and automata theory.          + rodger, s., et al. “engaging Middle
     Professor of the Practice                      algorithms, and parallel algorithms.          available via her Web site, the software         school teachers and students with alice
     of Computer science                                                                          is frequently downloaded and used by             in a diverse set of subjects.” accepted
                                                    honors and awards:                            universities around the world. she is            to 40th SIGCSE Technical Symposium on
                                                    aCM distinguished Member, 2006.               currently working on an evaluation study         Computer Science Education, 2009.
                                                                                                  of JFLaP involving faculty adopters at ten     + Horwitz, s., rodger, s., et al. “using
                                                                                                  other institutions.                              Peer-Led team Learning to increase
                                                                                                    Professor rodger developed JaWaa, an           Participation and success of under-
                                                                                                  easy-to-use, architecture independent            represented Groups in introductory
                                                                                                  tool that animates computer algorithms.          Computer science.” accepted to 40th
                                                                                                  Written in Java, the program provides an         SIGCSE Technical Symposium on Computer
                                                                                                  interface for users to write animations and      Science Education, 2009.
                                                                                                  then display them with any Web browser         + rodger, s., and Finley, t. JFLAP - An
                                                                                                  that supports Java. the animations               Interactive Formal Languages and
                                                                                                  are written by users in a simple script          Automata Package. Jones and Bartlett:
                                                                                                  language that can easily be learned by           sudbury, Ma, 2006.
                                                                                                  beginners with little or no programming
                                                                                                  experience. With the program, educators
                                                                                                  can quickly design animations relating to
                                                                                                  graphs, trees, stacks, queues, and arrays.
                                                                                                    Professor rodger created the duke
                                                                                                  emerging scholars program in Computer
                                                                                                  science (des-Cs) with startup funding
                                                                                                  from the national science Foundation.
                                                                                                  this program is for first-year students with
58
education faculty : dePartMent OF COMPuter sCienCe at duke university
s e c o n d a r y f a c u lt y                                                                 e m e r i t u s f a c u lt y


timothy lenoir                                                                                 dietolf ramm
kimberly Jenkins Professor                                                                     Professor emeritus of
for new technologies and society                                                               Computer science
Professor of Computer science


education:                                     dibner and alfred P. sloan Foundations,         education:
Ph.d., indiana university, 1974                and “How they Got Game,” a history of           Ph.d., duke university, 1969
B.a., st. Mary’s College, 1970                 interactive simulation and video games.         B.a., Cornell university, 1964
                                               With economists nathan rosenberg,
Professor timothy Lenoir is the kimberly       Henry rowen, and Brent Goldfarb he has          dr. ramm played an instrumental role in
Jenkins Chair for new technologies             just completed a collaborative study            the development of the undergraduate
and society at duke university. He has         for stanford university on stanford’s           program of computer science at duke.
published several books and articles on        historical relationship to silicon valley       Originally involved with the design of the
the history of biomedical science from         entitled, inventing the entrepreneurial         undergraduate major when it was first
the nineteenth century to the present.         region: stanford and the Co-evolution of        launched in 1973, dr. ramm helped shape
His more recent work has focused on the        silicon valley. in support of these projects,   its development throughout the years.
introduction of computers into biomedical      Lenoir has developed software tools for         in the classroom, he taught introductory
research from the early 1960s to the           interactive web-based collaboration. in         courses for both computer science majors
present, particularly the development of       this connection he is currently engaged         and non-majors.
computer graphics, medical visualization       with colleagues at uC santa Barbara in            dr. ramm also worked with colleagues
technology, the development of virtual         developing the nsF-supported Center             in the education group to develop new
reality and its applications in surgery and    for nanotechnology in society, where he         teaching techniques and tools to help
other fields. Lenoir has also been engaged     contributes to the effort to document the       students learn about innovative computer
in constructing online digital libraries for   history, societal, and ethical implications     science research. Focused on learning
a number of projects, including an archive     of bionanotechnology.                           new paradigms for programming
on the history of silicon valley. two recent                                                   and teaching, his efforts made the
projects include a web documentary                                                             department’s introductory courses what
project to document the history of                                                             they are today—packed with appropriate
bioinformatics funded by the Bern                                                              and cutting-edge content.



                                                                                                                                                                                                  59
                                                                                                                              dePartMent OF COMPuter sCienCe at duke university   : education faculty
Department of Computer science
Duke university
Box 90129
Durham, nC 27708-0129

ph: 919.660.6500
fax: 919.660.6519

www.cs.duke.edu

								
To top