Towards Dynamic Algorithm Selection for Numerical Black-Box Optimization: Investigating BBOB as a Use Case

TL;DR

This paper explores dynamic algorithm selection in evolutionary computation, demonstrating that even a single switch between algorithms during optimization can improve performance, using BBOB benchmark data.

Contribution

It introduces a simplified dynamic algorithm selection approach focusing on switching algorithms, and shows its potential benefits using BBOB data.

Findings

Single-switch dynAS can improve optimization performance.

BBOB data is useful for studying dynamic algorithm selection.

Challenges in dynAS are identified and discussed.

Abstract

One of the most challenging problems in evolutionary computation is to select from its family of diverse solvers one that performs well on a given problem. This algorithm selection problem is complicated by the fact that different phases of the optimization process require different search behavior. While this can partly be controlled by the algorithm itself, there exist large differences between algorithm performance. It can therefore be beneficial to swap the configuration or even the entire algorithm during the run. Long deemed impractical, recent advances in Machine Learning and in exploratory landscape analysis give hope that this dynamic algorithm configuration~(dynAC) can eventually be solved by automatically trained configuration schedules. With this work we aim at promoting research on dynAC, by introducing a simpler variant that focuses only on switching between different algorithms, not configurations. Using the rich data from the Black Box Optimization Benchmark~(BBOB) platform, we show that even single-switch dynamic Algorithm selection (dynAS) can potentially result in significant performance gains. We also discuss key challenges in dynAS, and argue that the BBOB-framework can become a useful tool in overcoming these.