www.inf.brad.ac.ukresearchseminarsppt2.ppt by dffhrtcv3

VIEWS: 0 PAGES: 29

									Ubiquitous Optimisation
Making Optimisation Easier to Use



          Prof Peter Cowling



     http://www.mosaic.brad.ac.uk
    Optimisation in Decision Making
                                       Outcomes
            Uncontrollable
               factors
                                                  D
                                                  e
Current                           D4              s
situation                                         i
                             D3                   r
                       D2                         a
                                                  b
                     D1                           i
                                                  l
                                                  i
  Controllable                                    t
    factors                                       y
                        Modelling
        Reflection             Conceptual
                                 Model

                  Extraction                   Creation
                                     Testing
   Tangible
    system
                                               Model
•Ill-structured                             •Well-structured
•Complex                                    •Simple
•Abstract                                   •Concrete
               Optimisation           NP-
                                      hard




                        Operational
                        Research
Evolutionary
Algorithms
                                      Novel
                                      Ideas
                  Artificial
                  Intelligence
           Does it work?
• Oil companies could not survive without
  optimisation
• Manufacturing/transport/logistics/
  project management – productivity
  improvements in the £billions worldwide
• Widely and expensively used in finance
  and management consultancy
Ubiquitous?
              Beneficiaries
• Any manager or engineer and every decision
  could benefit from a system which brought
  useful and usable optimisation.
• Consider the proliferation of spreadsheet use
  among managers/ engineers.
• The potential productivity improvements are
  in the £00,000,000,000s – from improved
  resource usage, better market targetting,
  better financial management.
     Advances which may bring
    ubiquitous optimisation closer
•   Speech/gesture input/output
•   Intelligent, learning computers
•   Cognitive science advances
•   Ambient computing
•   Control/sensor technologies
•   Increased IT awareness among
    managers/engineers
            Angles of attack
• Hyperheuristics, Software Toolboxes
  – Reducing the effort and expertise to model and
    solve problems
• Human-computer interaction and cognitive
  science
  – Integrating human and artificial intelligence
• Dynamic Optimisation – Stability and Utility
  – Reacting to the dynamic nature of real problems
• Gaining real-world problem experience
 Hyperheuristics
       Hyperheuristic
Heuristic
Choice


         Low level L.L. Heuristic
         heuristics performance
     Solution perturbation
                                Solution
                                quality

           Problem
  Benefits of Hyperheuristics
• Low level heuristics easy to implement
• Objective measures may be easy to
  implement – they should be present to
  raise decision quality
• Rapid prototyping – time to first
  solution low
        Concrete example
• Organising meetings at a sales summit
• Low level heuristics:
  – Add meeting, delete meeting, swap
    meeting, add delegate, remove delegate,
    etc.
• Objectives:
  – Minimise delegates
  – Maximise supplier meetings
         Concrete Example
• Hyperheuristic based on the exponential
  smoothing forecast of performance,
  compared to simple restarting approaches
• Result: 99 delegates reduced to 72 delegates
  with improved schedule quality for both
  delegates and suppliers
• Compares favourably with bespoke
  metaheuristic (Simulated Annealing)
  approach
• Fast to implement and easy to modify
          Other applications
•   Timetabling mobile trainers
•   Nurse rostering
•   Scheduling project meetings
•   Examination timetabling
      Other Hyperheuristics
• Genetic Algorithms
  – Chromosomes represent sequences of low
    level heuristics
  – Evolutionary ability to cope with changing
    environments useful
• Forecasting approaches
• Genetic Programming approaches
• Artificial Neural Network approaches
Human-Computer Interaction
STARK diagrams
Representing constraints
            Room capacity
            violation




                            Period limit
                            violation
STARK – some results
 100


 90


 80


 70                                                                                                     STARK 1

                                                                                                        STARK 2
 60
                                                                                                        STARK 3
 50
                                                                                                        CON 1

 40                                                                                                     CON 2

 30                                                                                                     CON 3
       1       7        13        19        25        31        37        43        49        55
           4       10        16        22        28        34        40        46        52        58


       Elasped tim e
                HuSSH
• Allowing users to create their own
  heuristics “on the fly”
• Capturing and reusing successful
  heuristic approaches allows the decision
  maker to work at a higher level
• User empowerment and satisfaction is
  raised by these approaches
• Users can learn to use sophisticated
  tools
                   HuSSH sample result
            810        m   m            m    u-s             u-s          m Fig. 2b
                                                                               500

            800                                                                 450

                                                                                400
            790
                                                                                350
No. Exams




            780
                                                                                300




                                                                                      Penalty
            770                                                                 250

                                                                                200
            760                                          í
                                                                                150
            750
                                                                                100
            740
                                                                                50

            730                                                                 0
                  10       20      30       40      50         60   70   80   Exams
                                                                              90
  u-              Unsched-Sched.
                                                 Time (%)
  m               Manual
                                                                              Penalty
Dynamic Scheduling - steel
                        Using Agents
`        user                   User agent



                                             HSM Agent
                SY Agent                                           coils


    Slabyard                                           Hot Strip Mill



                    CC-1 Agent CC-2 Agent CC-3 Agent

                              Continuous
                                                                  Slabs
                               Casters
                Ladle
  Stability, Utility and Robustness
Utility ( Sstatic, Sdynamic, E, t) = F dynamic - Fstatic

Stability ( S static , S dynamic , E , t )   i 1 C i  C i
                                                      N         '


Robustness (S)= R .Utility - (1-R).Stability,
where R is a real valued weight in the range [0,1].
E is the real-time event.
            Schedule Repair
                      Delete the non-available
                                coils




            Remaining Scheduled
Processed
                   coils
  coils
                                                 Reoptimise considering
                                                  the unscheduled coils
                         Unscheduled coils
        Simulation Prototype



Prototype Developed for Simulation
                     Some Results
            700


            600


            500
Stability




            400


            300


            200


            100


             0
             -700    -600     -500     -400      -300    -200   -100    0
                                           Utility


            NOT     SR      CSR      OSR      HCSR      HOSR    PR     CR
              Case studies
•   SORTED – Nationwide building society
•   SteelPlanner – A.I. Systems BV
•   Inventory Management – Meads
•   Workforce Scheduling - BT
•   Electronics Assembly - Mion
•   Nurse rostering – several Belgian
    Hospitals
 Conclusion – Open Problems
• Optimisation can improve productivity
• Optimisation can be made easier to use and
  more applicable
• Needed:
  – Robust, widely applicable optimisation
    algorithms/heuristics
  – Modelling languages and software toolboxes
  – Champions and consultants
  – Better understanding of human problem solving
    for use in HCI
  – Higher levels of computer use and literacy

								
To top