CMU Artificial Intelligence Repository
SCS: C port of Goldberg's Simple Classifier System
areas/genetic/ga/systems/scs/
SCS (Simple Classifier System) is a C port of the system from
Appendix D of Goldberg's "Genetic Algorithms in Search,
Optimization, and Machine Learning".
The following is a short description on the operation of the simple
classifier system:
The classifier system randomly generates a starting population
of classifiers. The program then generates a random signal to
the environment (in this case the multiplexer). The multiplexer
responds with the environmental message. The program then attempts
to match the random signal with the classifiers and those that match
are allowed to bid. The strengths of the matching classifiers are
deducted by the amount that is to be bid specified by the CBID variable
and the effective bids of the classifiers are compared. The classifer
with the largest effective bid (the bid modified by EBID) wins and the
classifier message is compared with the environmental message. If the
messages match, the classifier is given a reward. If the bucket brigade
flag is activated, the classifier will post a reward to be given to
the last matching classifier. The genetic algorithm is activated at
the iteration specified by the variable GAPERIOD. Pairs of
classifiers are chosen randomly according to their strengths. Bad
performing classifiers are replaced according to their similarities
to the new offsprings. Data about the population and its parameters
is recorded in REPORT.OUT at periodic intervals.
Version: 19-AUG-93
Requires: C
CD-ROM: Prime Time Freeware for AI, Issue 1-1
Author(s): Erik Mayer
Keywords:
Authors!Mayer, C!Code, Genetic Algorithms, Goldberg, SCS
References:
David E. Goldberg, "Genetic Algorithms in Search, Optimization, and
Machine Learning", Appendix D, Addison-Wesley, Boston, MA, 1989.
Last Web update on Mon Feb 13 10:23:13 1995
AI.Repository@cs.cmu.edu