Download - Learning Classifier Systems
![Page 1: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/1.jpg)
Learning Classifier Systems
![Page 2: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/2.jpg)
Learning Classifier Systems (LCS)
• The system has three layers:– A performance system that interacts with
environment,– An apportionment of credit algorithm that rates
rules as to usefulness,– A rule discovery algorithm that generates
plausible new rules to replace less useful rules.
![Page 3: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/3.jpg)
Performance System Cycles• Message is posted in the message list from the
input interface.• Each rule is matched against the message list• All matching rules compete to post in the next
message list via bidding process; winning rule posts in the new message list
• The output interface checks the new message and produces an effector action.
• The new message list replaces the previous one.• Repeat.
![Page 4: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/4.jpg)
Overview of LCS
![Page 5: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/5.jpg)
Rule format
• Rule– Condition = {0,1,#}k
– Action = message to be posted in the message list– Strength = rule’s usefulness to the system
![Page 6: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/6.jpg)
kind ears num. of legs smart scream runaway kiss
Example (Wolf or Grandmother?)
teeth
1 0 1 1 # 1 1 0
0 1 0 0 # 0 0 1
1
1
Wolf
GrandMa
Encoding
![Page 7: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/7.jpg)
Matching
[M] Condition Action Strength
# 1 # # #010 100
0 # 0 # 0011 100
Message List
0 1 0 0
Condition Action Strength
# 1 # # #010 100
1 # 0 1 1### 50
0 # 0 # 0011 100
1 # # # 1010 1000
1 0 1 # 0111 1000
[N]
![Page 8: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/8.jpg)
Bidding Process
[M] Rule id
Condition Action Strength
r1 # 1 # # #010 100
r3 0 # 0 # 0011 100
β = 0.2
Bid(r1) = 0.2 × ¼ × 100 = 5Bid(r3) = 0.2 × ½ × 100 = 10
r3 posts its message in the new message list.
Bid(R,t) = β × specificity(R) × Strength(R,t)Specificity(R)= number of non # / k
![Page 9: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/9.jpg)
Credit assignment: Bucket Brigade
r3Bucket
10
r5Bucket
150
coupled
Environment
executedReward
200
![Page 10: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/10.jpg)
r3Bucket
10
r5Bucket
150
EnvironmentReward
200
Credit assignment: Bucket Brigade
![Page 11: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/11.jpg)
Genetic Algorithms
• Fitness = rule strength• Parents: Strong classifiers (best, roulette
wheel, etc.)• Mutation: alter parts of parent’s string• Crossover: exchange parts of parents’ strings• Offspring replaces a weak rule.
![Page 12: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/12.jpg)
Genetic Algorithms (cont.)
0 0 1 0 1 1 # #
1 0 1 0 0 1 0 0
Parent 1Parent 2
0 0 1 0 1 1 0 0
1 0 1 0 0 1 # #
Crossover point
0 0 1 0 1 1 # #
1 0 1 0 0 1 0 0
Parent 1Parent 2
0 0 1 0 1 0 # #
1 0 1 0 0 1 0 0
Crossover
Mutation
![Page 13: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/13.jpg)
Maze Environment
A
Environment
Message List
40 5 f N 5 (1,2)
GF
Condition Action Strength
# >0 # # # # GF 1000
# <0 # # # # ∧ TL TL 1000
# <0 # # # # ∧ TR TR 1000
(Signal smell-ahead bump heading score location)
![Page 14: Learning Classifier Systems](https://reader035.vdocument.in/reader035/viewer/2022062304/56813669550346895d9df60f/html5/thumbnails/14.jpg)
References
• A Mathematical framework for Studying Learning in Classifier Systems, John H. Holland, Phsyca D, Vol 2, No 1-3, 1986, pp. 307-317
• A First Order Logic Classifier System, Drew Mellor Gecco ’05