应用于锥形约束问题的带投影修复的 $(\mu/\mu_I,\lambda)$-$\sigma$-自适应演化策略分析
神经与进化计算
2018-12-18 v1
摘要
针对一个锥形约束问题，给出了 --自适应演化策略（SA-ES）的理论性能分析。不可行后代通过投影到可行域边界上进行修复。采用闭式近似描述演化策略的单代进展。构建了近似确定性演化方程以分析策略的动态。通过用近似单代表达式迭代演化方程，可预测演化策略的动态。将推导的理论结果与实验进行比较以评估近似质量。结果表明，在稳态下 -SA-ES 表现出如同该 ES 在优化球面模型时的性能。与非重组的 -ES 不同，父代稳态行为并不在锥边界上演化，而是在一定程度上远离边界。
引用
@article{arxiv.1812.06300,
title = {Analysis of the $(\mu/\mu_I,\lambda)$-$\sigma$-Self-Adaptation Evolution Strategy with Repair by Projection Applied to a Conically Constrained Problem},
author = {Patrick Spettel and Hans-Georg Beyer},
journal= {arXiv preprint arXiv:1812.06300},
year = {2018}
}
备注
This is a PREPRINT of an article submitted to IEEE Transactions On Evolutionary Computation. It is currently under review. Due to size limitations, this manuscript comprises figures with reduced resolution. 10 pages + supplementary material. Copyright 2018 IEEE. The work was supported by the Austrian Science Fund FWF under grant P29651-N32