Department
Computer Science
Major
Computer Science
Research Advisor
Tauritz, Daniel
Advisor's Department
Computer Science
Funding Source
UMR Opportunities for Undergraduate Research Experiences (OURE) Program
Abstract
One of the primary obstacles to Evolutionary Algorithms (EAs) fulfilling their promise as easy to use general-purpose problem solvers is the difficulty of correctly configuring them for specific problems such as to obtain satisfactory performance. This paper introduces the concept of democratic, semiautonomous parent selection by encoding and evolving population rating operators as in Genetic Programming and shows the potential of extending self-adaptation by pairing mates using an adaptation of the Stable Roommates problem. Replacing the typical general parent selection algorithm with autonomously evolved individual selection parameters has the prospective to bring EAs a step closer to their promise as easy to use general-purpose problem solvers.
Biography
Josh M. Eads is a junior at the University of Missouri - Rolla majoring in Computer Science and Applied Mathematics. He is actively involved with Rolla's ACM chapter and currently working on research involving Genetic Programming and Evolutionary Algorithms. Josh plans to continue his education and research as a graduate student after completing his degrees in Computer Science and Math.
Research Category
Engineering
Presentation Type
Poster Presentation
Document Type
Poster
Location
Havener Center, Carver-Turner Room
Presentation Date
11 April 2007, 9:00 am - 11:45 am
Self-Adaptive Semi-Autonomous Democratic Parent Selection
Havener Center, Carver-Turner Room
One of the primary obstacles to Evolutionary Algorithms (EAs) fulfilling their promise as easy to use general-purpose problem solvers is the difficulty of correctly configuring them for specific problems such as to obtain satisfactory performance. This paper introduces the concept of democratic, semiautonomous parent selection by encoding and evolving population rating operators as in Genetic Programming and shows the potential of extending self-adaptation by pairing mates using an adaptation of the Stable Roommates problem. Replacing the typical general parent selection algorithm with autonomously evolved individual selection parameters has the prospective to bring EAs a step closer to their promise as easy to use general-purpose problem solvers.
