引 言 (1)
• DARPA, 2003, PAL(Perceptive Assistant Learns, 2003~2008)
• SRI, CALO(Cognitive Assistant that Learns and Organizes)
Latin word "calonis", which means "soldier’s servant".
• The goal of the project is to create cognitive software
systems, that is, systems that can reason, learn from
experience, be told what to do, explain what they are
doing, reflect on their experience, and respond robustly
引 言 (2)
• Artificial Intelligence, Machine Learning, Natural Language
Processing, Knowledge Representation, Human-computer
Interaction, Flexible Planning, and Behavioral Studies
• 美国斯坦福国际研究院（Stanford Research Institute International，
• HTTP://www.ai.sri.com/project/CALO, HTTP://caloproject.sri.com/
• 22家研究机构, 250科研人员
引 言 (3)
引 言 (4)
ORGANIZE AND MANAGE INFORMATION
PREPARE INFORMATION PRODUCTS
OBSERVE AND MEDIATE INTERACTIONS
MONITOR AND MANAGE TASKS
SCHEDULE AND ORGANIZE IN TIME
ACQUIRE AND ALLOCATE RESOURCES
OAA (Open Agent Architecture) http://www.ai.sri.com/oaa/
• Perrault通过麦克风通知CALO系统: 当关于安全的邮件到达时立
• Cheyer写了一封标题为“security alert”的邮件给Perrault;
Collaboration Process (1)
Collaboration Process (2)
Collaboration Process (3)
Collaboration Process (4)
Collaboration Process (5)
Collaboration Process (6)
Characteristics [Martin, AAI99] [Cheyer, AAMAS01]
agents can be created in many languages and interface with existing
agents can be added or replaced on the fly
• User friendly
high-level, natural expression of delegated tasks
• Developer friendly
Unified approach to service provision, data management, and task
handwriting, speech, gestures, and direct manipulation can be
20 Unanticipated sharing across many applications
ICL (Interagent Communication Language)
• A layer of conversational protocol defined by event types, similar
• A content layer consists of the specific goals, triggers, and data
elements, similar with KIF.
• Based on an extension of the Prolog language.
• All communications between agents occur in the form of events.
Provide a general mechanism for specifying some action to be
taken when some set of conditions is met.
• Delegation, optimization, interpretation
Declarations of solvables
• solvable(GoalTemplate, Parameters, Permissions)
• solvable(send_message(email, +ToPerson, +Params),
[type(procedure), callback(send_mail)], )
• solvable(last_message(email, -MessageId), [type(data),
SPARK (SRI Procedural Agent Realization Kit)
• PRS, and shares the same Belief Desire Intention (BDI)
model of rationality.
• Support the construction of large-scale, practical agent
systems, and contains sophisticated mechanisms for
encoding and controlling agent behavior.
• Has a well-defined semantic model that is intended to
There is a need for agent systems that can scale to real world applications,
support reasoning about the agents' knowledge and
yet retain the clean semantic underpinning of more formal agent frameworks.
AAMAS04] [Morley, AAAI04]
Overall Architecture for a SPARK Agent
• A Knowledge base of beliefs about the world and itself that is
updated both by sensory input from the external world and by
• provide declarative representations of activities for
responding to events and for decomposing complex tasks
into simpler tasks.
• At any given time the agent has a set of intentions, which are
procedure instances that it is currently executing.
• Is SPARK’s core. Its role is to manage the execution of
• It does this by repeatedly selecting one of the current
intentions to process and performing a single step of that
• Steps generally involve activities such as performing tests on
and changing the KB, adding tasks, decomposing tasks by
applying procedures, or executing primitive actions.
IRIS: Integrate. Relate. Infer. Share.
• Semantic Desktop [Cheyer, Semantic Web05]
• CALO is an artificial intelligence application for which
IRIS serves as the semantic desktop user interface.
• Information resources
• A knowledge base
• User interface framework
• IRIS is used to semantically integrate the tools of
• Clib (the Component Library Specification)
Consists of definitions for everyday objects and
Use OWL as the data representation.
• One of the key differentiators of IRIS, compared to
many semantic desktop systems, is the emphasis on
machine learning and the implementation of a plug-and-
play learning framework.
• A typical use case
Learn from Files.
Classification According to Project.
30 Higher-level Reasoning
• Shared structures are essential for both end-user
applications, such as team decision making and project
• and for infrastructural components such as machine
learning algorithms, which improve when given larger
data sets to work on.
PTIME (Personalized Time Management) [Berry,
• PTIME will unobtrusively learn user preferences through a
combination of passive learning, active learning, and advice-
• As above result, over time the user will become more
confident of PTIME’s ability, and will thus let it make more
• And as autonomy increases, PTIME will learn when to
involve the user in its decisions.
Three components of PTIME
• Process Controller (Heart of PTIME)
A SPARK agent that captures possible interactions.
Manages PTIME’s processes, tasking and coordinating the
activities of the Constraint Reasoner and Preference Learner.
• Constraint Reasoner
Explore conflict resolution options using relaxation, event
bumping, and explanation techniques.
• Preference Learner
Is an unobtrusive, online learner where the user’s selections
from suggested alternatives provide feedback to the learning
Research Directions [Berry, AAAI05]
• Soft CSP design [Venable, IJCAI05]
Simple Temporal Problem (STP)
Disjunctive Temporal Problem (DTP)
Simple Temporal Problem with Uncertainty (STPU)
Disjunctive Temporal Problem with Uncertainty
• Negotiation: Process Design for Conflict Resolution
• Learning for Adjustable Autonomy
SR/AR (Situation Assessment / Activity Recognition)
• Empower CALO with the ability to interpret and make
sense of what is going on in its environment.
• Large, dynamic and relational state space.
• Large sources of temproal and multi-model data.
• Semantic gaps, uncertainty.
• T1: Methods for state estimation in relational domains, including
dealing with unknown number of objects and their identity,
relevance determination and focus of attention.
• T2: Methods for inference and learning in continuous time complex
• T3: Methods for active learning, strategic user querying and fast
inference in large HMM.
• T4: Methods for learning and recognizing hierarchical activity
models from desktop activity traces.
• T5: Methods for location-based activity recognition.
• T6: Methods for learning and recognizing activities, gestures and
relevant objects from low-level physical sensors.
• T7: Methods for state estimation in communicative activities.
• T8: Methods for tracking the progress of the CALO plan, including
possible failures and missed deadlines.
T1: Situation assessment in relational domain
• Develop a language for representing domain in which the number
of objects and their identity is unknown ---- BLOG (Bayesian LOGic)
and DBLOG (Dynamic BLOG).
• Propose an approach based on probabilistic relational models that
does not insist on making a complete propositionalization of the
domain at inference time.
T2: Continous time modeling in complex dynamic
• From DBN to CTBN (Continuous Time Bayesian Network).
T3: Active learning, strategic user querying, and fast
inference in large HMM
• Have implemented active learning for HMMs and obtained promising
results on user activity data from an instrumented desktop.
• Will extend these results to the domain of general graphical models,
T4: Learning and recognizing user’s activities from desktop
• Typical user’s activities have an inherent hierarchical structure.
• The main challenge for CALO is to chain the related events together, and
infer the hidden sub-activity and activity at the high-level.
• Efficient inference algorithms and semi-supervised learning
approach in abstract and hierarchical hidden Markov models, with
continuous time Bayesian network
T5: Location-based activity recognition
• Develop techniques that can reliably estimate the location (Location
information is extracted from WiFi signal strength).
• Develop methods for learning and inferring higher-level patterns of
movement and activities from the data generated by a location-
• From RMNs (Relational Markov Networks) to RFGs (Relational
T6, T7 and T8
• HHMM (Hierarchical Hidden Markov Models) [Nguyen, CVPR05]
• ProPL (Probabilistic Process Language)
Transfer Learning [Dietterich 05]
• Replacing an employee
Employee A is leaving an organization and being replaced by
employee B. Can B’s CALO demonstrate transfer based on
learning that took place in A’s CALO?
• Moving to a new job
An employee leaves organization A and moves to a new
organization B. Can his CALO demonstrate transfer learning
from experiences in A to capabilities in B?
Some learning mechanisms for transfer learning
• Hierarchical Bayesian learning
• Shared parameter models
• Instance weighting
• Abstraction regularization
• Cascading classifiers
• Attribute Weights and Low Dimensional
展望 (3-CALO Learning)
Relational: Learn relationships among entities
Category: Learn relevant groupings Sequential: Learn the dynamic structure of
for observed information ongoing activity of the user
Procedural: Learn to handle
Language: Learn new new tasks through planning
Information from text and Observation
Factual: Reason to learn
Inference new facts
Perceptual: Learn to associate images
Long-Term and sounds with other knowledge
Advice: Learn from
展望 (4-Using CALO Learning)
Learn when to Mary John Harry
Learn to handle
Associate people with Learn to adapt to
roles and places new situations
展望 (5-Technical Challenges)
Robust mixed-initiative multitasking
Enduring improvement in a changing environment
Integration of heterogeneous t
Establishing and maintaining trust
Seamless use across platforms
[Morley, AAMAS04] Morley, D. and Myers, K. The SPARK Agent
Framework. In Proc. of the Third Int. Joint Conf. on Autonomous
Agents and Multi Agent Systems (AAMAS-04), New York, NY, pp. 712-
719, July 2004.
[Morley, AAAI04] Morley, D. and Myers, K. Balancing Formal and
Practical Concerns in Agent Design. In Proc. of the AAAI Workshop on
Intelligent Agent Architectures: Combining the Strengths of Software
Engineering and Cognitive Systems, 2004.
[Cheyer, Semantic Web05] Cheyer, A. and Park, J. and Giuli, R. IRIS:
Integrate. Relate. Infer. Share. In 1st Workshop on The Semantic
Desktop. 4th International Semantic Web Conference, p. 15, Nov 2005.
[Berry, AAMAS06] Berry, P. and Conley, K. and Gervasio, M. and
Peintner, B. and Uribe, T. and Yorke-Smith, N. Deploying a
Personalized Time Management Agent, in Proceedings of the Fifth
International Joint Conference on Autonomous Agents and Multi Agent
50 Systems (AAMAS’06) Industrial Track, Hakodate, Japan, May 2006.
[Berry, AAAI05] Berry, P. and Gervasio, M. and Uribe, T. and Pollack,
M. and Moffitt, M. A Personalized Time Management Assistant, in AAAI
2005 Spring Symposium Series, Stanford, CA, Mar 2005.
[Venable, IJCAI05] Venable, K. B. and Yorke-Smith, N. Disjunctive
Temporal Planning with Uncertainty, in Proceedings of Nineteenth
International Joint Conference on Artificial Intelligence (IJCAI’05),
Edinburgh, UK, pp. 1385–1386, Aug 2005.
[Nguyen, CVPR05] Nguyen, N. and Phung, D. and Venkatesh, S. and
Bui, H. Learning and detecting activities from movement trajectories
using the hierarchical hidden Markov model, in IEEE International
Conference on Computer Vision and Pattern Recognition, 2005.
[Duong, CVPR05] Duong, T. and Bui, H. and Phung, D. and Vekatesh,
S. Activity recognition and abnormality detecting with the switching
hidden semi-Markov model, in IEEE International Conference on
Computer Vision and Pattern Recognition, 2005.
[Hung 05] Hung Bui. Situation Assessment and Activity Recognition.
Technique Report, SRI International, 2005.
[Dietterich 05] Tom Dietterich, Girish Acharya. Transfer Learning
Activity for Years 3-5. Technique Report, SRI International, 2005.
[Martin, AAI99] Martin, David L. and Cheyer, Adam J. and Moran,
Douglas B. The Open Agent Architecture: A Framework for Building
Distributed Software Systems. Applied Artificial Intelligence, vol. 13, no.
1-2, pp. 91-128, January-March 1999.
[Cheyer, AAMAS01] Cheyer, Adam and Martin, David. The Open
Agent Architecture. Journal of Autonomous Agents and Multi-Agent
Systems, vol. 4 , no. 1, pp. 143-148, March 2001.