AI SearchClassifier systems and genetic algorithms : a case study in bargaining
Magcamit, Domingo
Classifier systems and genetic algorithms : a case study in bargaining - Bangkok : Asian Institute of Technology, 1991 - 70 p. - Thesis ; no. CS-91-2 . - Asian Institute of Technology. Thesis ; no. CS-91-2 .
A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science, School of Engineering and Technology
Thesis (M.Sc.) - Asian Institute of Technology, 1991
The performance of a classifier system which belongs to the genetics based machine learning architecture is tested among ten different options utilizing the various genetic algorithm (GA) operators, namely : reproduction, crossover, inversion and mutation. Together with these GA operators, apportionment of credit works hand in hand to determine the fittest set of string rules expressed as classifiers which serves as solution set in a two-party bargaining problem. Two related models are run adopting decimal coding in the first model and binary representation in the second one. The concept of schemata is incorporated in both models and results from both conform with each other. To search for convergence, classifiers' fitness, eagerness and specificity are compared and patterned after the way human behavior is observed while under negotiation. The system's capability to 'learn' rules is explored by selecting the best string rules that provide solution options to the user. Similar with the common negotiation support system and multicriteria decision making (MCDM) approach, the system offers alternative solutions, but the final decision always resting on the decision maker.
Negotiation--Decision-making
Algorithms
Classifier systems and genetic algorithms : a case study in bargaining - Bangkok : Asian Institute of Technology, 1991 - 70 p. - Thesis ; no. CS-91-2 . - Asian Institute of Technology. Thesis ; no. CS-91-2 .
A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science, School of Engineering and Technology
Thesis (M.Sc.) - Asian Institute of Technology, 1991
The performance of a classifier system which belongs to the genetics based machine learning architecture is tested among ten different options utilizing the various genetic algorithm (GA) operators, namely : reproduction, crossover, inversion and mutation. Together with these GA operators, apportionment of credit works hand in hand to determine the fittest set of string rules expressed as classifiers which serves as solution set in a two-party bargaining problem. Two related models are run adopting decimal coding in the first model and binary representation in the second one. The concept of schemata is incorporated in both models and results from both conform with each other. To search for convergence, classifiers' fitness, eagerness and specificity are compared and patterned after the way human behavior is observed while under negotiation. The system's capability to 'learn' rules is explored by selecting the best string rules that provide solution options to the user. Similar with the common negotiation support system and multicriteria decision making (MCDM) approach, the system offers alternative solutions, but the final decision always resting on the decision maker.
Negotiation--Decision-making
Algorithms