Pareto-based evolutionary multiobjective approaches and the generalized Nash equilibrium problem
| Autor: | Mihai Alexandru Suciu, Rodica Ioana Lung, Noémi Gaskó |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
TheoryofComputation_MISCELLANEOUS
Computer Science::Computer Science and Game Theory Mathematical optimization Control and Optimization Relation (database) Computer Networks and Communications Computer science Computation 0211 other engineering and technologies Evolutionary algorithm 02 engineering and technology Management Science and Operations Research Multi-objective optimization Set (abstract data type) symbols.namesake Artificial Intelligence 0202 electrical engineering electronic engineering information engineering 021103 operations research Pareto principle TheoryofComputation_GENERAL Nash equilibrium Dominance (economics) symbols 020201 artificial intelligence & image processing Software Information Systems |
| Zdroj: | Journal of Heuristics. 26:561-584 |
| ISSN: | 1572-9397 1381-1231 |
| DOI: | 10.1007/s10732-020-09438-w |
| Popis: | Pareto-based evolutionary multiobjective approaches are methods that use the Pareto dominance concept to guide the search of evolutionary algorithms towards the Pareto frontier of a problem. To address the challenge of providing an entire set of optimal solutions they use specially designed mechanisms for preserving search diversity and maintaining the non-dominated solutions set. The limitation of the Pareto dominance relation in high-dimensional spaces has rendered these methods inefficient for many-objective optimization. In this paper we aim to exploit existing Pareto-based methods to compute the generalized Nash equilibrium for multi-player games by replacing the Pareto dominance relation with an equilibrium generative relation. The generalized Nash equilibrium extends the Nash equilibrium concept by considering constraints over players’ strategies. Numerical experiments indicate that the selected methods can be employed for equilibria computation even for games with up to twenty players. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink