case injected genetic algorithms
DESCRIPTION
Case Injected Genetic Algorithms. Sushil J. Louis Genetic Algorithm Systems Lab (gaslab) University of Nevada, Reno http://www.cs.unr.edu/~sushil http://gaslab.cs.unr.edu/ [email protected]. Learning from Experience: Case Injected Genetic Algorithm Design of Combinational Logic Circuits. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/1.jpg)
Case Injected Genetic Algorithms
Sushil J. LouisGenetic Algorithm Systems Lab (gaslab)
University of Nevada, Reno
http://www.cs.unr.edu/~sushil
http://gaslab.cs.unr.edu/
![Page 2: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/2.jpg)
Learning from Experience: Case Injected Genetic Algorithm Design of Combinational Logic Circuits
Sushil J. LouisGenetic Algorithm Systems Lab (gaslab)
University of Nevada, Reno
http://www.cs.unr.edu/~sushil
http://gaslab.cs.unr.edu/
![Page 3: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/3.jpg)
http://gaslab.cs.unr.edu
Outline
Motivation What is the technique?
Genetic Algorithm and Case-Based Reasoning
Is it useful? Evaluate performance on Combinational Logic Design
Results Conclusions
![Page 4: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/4.jpg)
http://gaslab.cs.unr.edu
Outline
Motivation What is the technique?
Genetic Algorithm and Case-Based Reasoning Is it useful?
Combinational Logic Design Strike Force Asset Allocation TSP Scheduling
Conclusions
![Page 5: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/5.jpg)
http://gaslab.cs.unr.edu
Genetic Algorithm
Non-Deterministic, Parallel, Search Poorly understood problems Evaluate, Select, Recombine Population search
Population member encodes candidate solution Building blocks combine to make progress More resistant to local optima Iterative, requiring many evaluations
![Page 6: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/6.jpg)
http://gaslab.cs.unr.edu
Motivation
Deployed systems are expected to confront and solve many problems over their lifetime
How can we increase genetic algorithm performance with experience?
Provide GA with a memory
![Page 7: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/7.jpg)
http://gaslab.cs.unr.edu
Case-Based Reasoning
When confronted by a new problem, adapt similar (already solved) problem’s solution to solve new problem
CBR Associative Memory + Adaptation CBR: Indexing (on problem similarity) and
adaptation are domain dependent
![Page 8: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/8.jpg)
http://gaslab.cs.unr.edu
Case Injected Genetic AlgoRithm
Combine genetic search with case-based reasoning
Case-base provides memory Genetic algorithm provides adaptation Genetic algorithm generates cases
Any member of the GA’s population is a case
![Page 9: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/9.jpg)
http://gaslab.cs.unr.edu
System
![Page 10: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/10.jpg)
http://gaslab.cs.unr.edu
Related work
Seeding:Koza, Greffensttette, Ramsey, Louis Lifelong learning: Thrun Key Differences
Store and reuse intermediate solutions Solve sequences of similar problems
![Page 11: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/11.jpg)
http://gaslab.cs.unr.edu
Combinational Logic Design
An example of configuration design Given a function and a target technology to
work with design an artifact that performs this function subject to constraints Target technology: Logic gates Function: Parity checking Constraints: 2-D gate array
![Page 12: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/12.jpg)
http://gaslab.cs.unr.edu
Encoding
![Page 13: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/13.jpg)
http://gaslab.cs.unr.edu
Encoding
![Page 14: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/14.jpg)
http://gaslab.cs.unr.edu
Parity
Input 3-bit Parity 3-1 problem000 0 0
001 1 0
010 1 1
011 0 0
100 1 1
101 0 0
110 0 0
111 1 1
![Page 15: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/15.jpg)
http://gaslab.cs.unr.edu
Which cases to inject?
Problem distance metric (Louis ‘97) Domain dependent
Solution distance metric Genetic algorithm encodings
Binary – hamming distanceReal – euclidean distancePermutation – longest common substring…
![Page 16: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/16.jpg)
http://gaslab.cs.unr.edu
Problem similarity
![Page 17: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/17.jpg)
http://gaslab.cs.unr.edu
Lessons
Storing and Injecting solutions may not improve solution quality
Storing and Injecting partial solutions does lead to improved quality
![Page 18: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/18.jpg)
http://gaslab.cs.unr.edu
OSSP Performance
![Page 19: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/19.jpg)
http://gaslab.cs.unr.edu
Solution Similarity
![Page 20: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/20.jpg)
http://gaslab.cs.unr.edu
Periodic Injection Strategies
Closest to best Furthest from worst Probabilistic closest to best Probabilistic furthest from worst Randomly choose a case from case-base Create random individual
![Page 21: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/21.jpg)
http://gaslab.cs.unr.edu
Setup
50, 6-bit combinational logic design problems
Randomly select and flip bits in parity output to define logic function
Compare performance Quality of final design solution (correct output) Time to this final solution (in generations)
![Page 22: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/22.jpg)
http://gaslab.cs.unr.edu
Parameters
Population size: 30 No of generations: 30 CHC (elitist) selection Scaling factor: 1.05 Prob. Crossover: 0.95 Prob. Mutation: 0.05
Store best individual every generation
Inject every 5 generations (2^5 = 32)
Inject 3 cases (10%) Multiple injection
strategies
Averages over 10 runs
![Page 23: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/23.jpg)
http://gaslab.cs.unr.edu
Problem distribution
![Page 24: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/24.jpg)
http://gaslab.cs.unr.edu
Performance - Quality
![Page 25: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/25.jpg)
http://gaslab.cs.unr.edu
Performance - Time
![Page 26: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/26.jpg)
http://gaslab.cs.unr.edu
Injection Strategies
![Page 27: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/27.jpg)
http://gaslab.cs.unr.edu
Solution distribution
![Page 28: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/28.jpg)
http://gaslab.cs.unr.edu
Strike force asset allocation
Allocate platforms to targets
Dynamic Changing Priority Battlefield conditions Popup Weather …
![Page 29: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/29.jpg)
http://gaslab.cs.unr.edu
Factors in allocation
Pilot proficiency Asset suitability Priority Risk
Route Other assets (SEAD) Weather
![Page 30: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/30.jpg)
http://gaslab.cs.unr.edu
Maximize mission success
Binary encoding Platform to multiple targets Target can have multiple platforms Dynamic battle-space
Strong time constraints
![Page 31: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/31.jpg)
http://gaslab.cs.unr.edu
Setup
50 problems. 10 platforms, 40 assets, 10 targets Each platform could be allocated to two targets Problems varied in risk matrix Popsize=80, Generations=80, Pc=1.0, Pm=0.05,
probabilistic closest to best, injection period=9, injection % = 10% of popsize
![Page 32: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/32.jpg)
http://gaslab.cs.unr.edu
Results
![Page 33: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/33.jpg)
http://gaslab.cs.unr.edu
TSP
Find the shortest route that visits every city exactly once (except for start city)
Permutation encoding. Ex: 35412 Similarity metric: Longest common
subsequence (Cormen et al, Introduction to Algorithms)
50 problems, move city locations
![Page 34: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/34.jpg)
http://gaslab.cs.unr.edu
TSP performance
![Page 35: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/35.jpg)
http://gaslab.cs.unr.edu
Scheduling
Job shop scheduling problems Permutation encoding (Fang) Similarity metric: Longest common
subsequence (Cormen et al, Introduction to Algorithms)
50 problems, change task lengths
![Page 36: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/36.jpg)
http://gaslab.cs.unr.edu
JSSP Performance (10x10)
![Page 37: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/37.jpg)
http://gaslab.cs.unr.edu
JSSP Performance (15x15)
![Page 38: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/38.jpg)
http://gaslab.cs.unr.edu
Summary
Case Injected Genetic AlgoRithm: A hybrid system that combines genetic algorithms with a case-based memory
Defined problem-similarity and solution-similarity metrics
Defined performance metrics and showed empirically that CIGAR learns to increase performance for sequences of similar problems
![Page 39: Case Injected Genetic Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062409/5681470b550346895db44422/html5/thumbnails/39.jpg)
http://gaslab.cs.unr.edu
Conclusions
Case Injected Genetic AlgoRithm is a viable system for increasing performance with experience
Implications for system design Increases performance with experience Generates cases during problem solving Long term navigable store of expertise Design analysis by analyzing case-base