On Automated Discovery of Models Using Genetic Programming: Bargaining in a Three-Agent Coalitions Game

Abstract:The creation of mathematical, as well as qualitative (or rule-based), models is difficult, time-consuming, and expensive. Recent developments in evolutionary computation hold out the prospect that, for many problems of practical import, machine learning techniques can be used to discover useful models automatically. The prospects are particularly bright, we believe, for such automated discoveries in the context of game theory. This paper reports on a series of successful experiments in which we used a genetic programming regime to discover high-quality negotiation policies. The game-theoretic context in which we conducted these experiments—a three-player coalitions game with sidepayments—is considerably more complex and subtle than any reported in the previous literature on machine learning applied to game theory.Key Words and Phrases:: automatic model discoverygame theorygenetic programmingmachine learning Additional informationNotes on contributorsGarett DwormanGarett Dworman is a Ph.D. candidate in the Department of Operations and Information Management at the Wharton School of the University of Pennsylvania. He studies the application of cognitively motivated information systems to decision problems. Two current research projects are (1) investigations into computational rationality in bargaining situations using evolutionary computation (e.g., genetic algorithms and genetic programming), and (2) developing information retrieval systems capable of high-level pattern discovery in document collections.Steven O. KimbroughSteven O. Kimbrough is an Associate Professor at the Wharton School of the University of Pennsylvania. His main research interests are in the fields of decision support and artificial intelligence, logic modeling, the computational theory of rationality, and electronic commerce. His active research areas include computational approaches to belief revision and nonmonotonic reasoning, formal languages for business communication, and knowledge-based decision support systems.James D. LaingJames D. Laing is a member of the Decision Processes and the Information Systems faculty at the Wharton School of the University of Pennsylvania. His research interests include cooperative and noncooperative game theory, process models of nonrational behavior, statistical methods, laboratory studies, and evolutionary computation, with particular emphasis on applications in the science of multilateral negotiations.

• 저널Journal of Management Information Systems · 12(3) · 97–125
• 토픽Game Theory and Applications · Management Science and Operations Research
• DOI10.1080/07421222.1995.11518093
• 저자Garett Dworman, Steven O. Kimbrough, James D. Laing