[1J5-GS-2] Machine learning: Fundamental theory (2)

[1J5-GS-2-04] CMA-ES with Coordinate Selection Based on Curvature Estimation for High Dimensional Black-box Optimization

〇Hiroki Shimizu¹, Masashi Toyoda² (1. The University of Tokyo, 2. The Institute of Industrial Science, The University of Tokyo)

Keywords:Black-box optimization, Evolutionary computation, Machine learning

We propose a coordinate selection technique for the covariance matrix adaptation evolution strategy (CMA-ES).
Our technique enables CMA-ES to adapt to objective functions with ill-conditionality and deep dependencies of parameters based on curvature estimation.
We examine properties of our technique through several benchmark functions.