Saving Resources with Plagues in Genetic Algorithms
The population size of genetic algorithms (GAs) affects the quality of the solutions and the time required to find them. While progress has been made in estimating the population sizes required to reach a desired solution quality for certain problems, in practice the sizing of populations is still usually performed by trial and error. These trials might lead to find a population that is large enough to reach a satisfactory solution, but there may still be opportunities to optimize the computational cost by reducing the size of the population. This paper presents a technique called plague that periodically removes a number of individuals from the population as the GA executes. Recently, the usefulness of the plague has been demonstrated for genetic programming. The objective of this paper is to extend the study of plagues to genetic algorithms. We experiment with deceptive trap functions, a tunable difficult problem for GAs, and the experiments show that plagues can save computational time while maintaining solution quality and reliability.
- Research Organization:
- Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 15014317
- Report Number(s):
- UCRL-CONF-204742; TRN: US200803%%587
- Resource Relation:
- Journal Volume: 3242; Conference: Presented at: Parallel Problem Solving from Nature, Birmingham, United Kingdom, Sep 18 - Sep 22, 2004
- Country of Publication:
- United States
- Language:
- English
Similar Records
A genetic algorithm solution to the unit commitment problem
Economic Dispatch Using Genetic Algorithm Based Hybrid Approach