Update-Based Evolution Control
Haiping Ma, Minrui Fei, Dan Simon, and Hongwei Mo
Abstract: Evolutionary algorithms (EAs) are robust optimization methods that have been used in many engineering applications. However, real-world fitness evaluations can be computationally expensive, so it may be necessary to estimate the fitness with an approximate model. This paper reviews design and analysis of computer experiments (DACE) as an approximation method that combines a global polynomial with a local Gaussian model to estimate continuous fitness functions. The paper incorporates DACE in various EAs, including a genetic algorithm (GA), population-based incremental learning (PBIL), differential evolution (DE), and particle swarm optimization (PSO), to test unconstrained and constrained benchmark problems, both with and without fitness function evaluation noise. The paper also introduces a new evolution control strategy called update-based control that estimates the fitness of certain individuals each generation based on the exact fitness values of other individuals during that same generation. Results show that update-based evolution control outperforms other strategies on noise-free, noisy, constrained, and unconstrained benchmarks. The results also show that update-based evolution control can compensate for fitness evaluation noise.
- H. Ma, M. Fei, D. Simon, and H. Mo, "Update-based evolution control: A new fitness approximation method for evolutionary algorithms," Engineering Optimization, vol. 47, no. 9, pp. 1177-1190, 2015 - pdf, 297 KB
- Supplemental Data - pdf, 689 KB - This file includes several tables and figures that explain the material in the above-referenced paper, but that were too extensive to include in the published paper.
Last Revised: June 18, 2015