Asynchronous \epsilon-Greedy Bayesian Optimisation

TL;DR

This paper introduces AEGiS, an asynchronous Bayesian optimisation method that combines greedy, Thompson sampling, and random strategies to efficiently optimise expensive black-box functions, outperforming existing methods.

Contribution

The paper proposes AEGiS, a novel asynchronous BO algorithm that balances exploration and exploitation using an epsilon-greedy approach, improving resource utilization and optimisation performance.

Findings

AEGiS outperforms existing asynchronous BO methods on benchmarks.

Performance of AEGiS is comparable to expected improvement when using a single worker.

Empirical results show AEGiS's effectiveness on real-world problems.

Abstract

Batch Bayesian optimisation (BO) is a successful technique for the optimisation of expensive black-box functions. Asynchronous BO can reduce wallclock time by starting a new evaluation as soon as another finishes, thus maximising resource utilisation. To maximise resource allocation, we develop a novel asynchronous BO method, AEGiS (Asynchronous $\epsilon$-Greedy Global Search) that combines greedy search, exploiting the surrogate's mean prediction, with Thompson sampling and random selection from the approximate Pareto set describing the trade-off between exploitation (surrogate mean prediction) and exploration (surrogate posterior variance). We demonstrate empirically the efficacy of AEGiS on synthetic benchmark problems, meta-surrogate hyperparameter tuning problems and real-world problems, showing that AEGiS generally outperforms existing methods for asynchronous BO. When a single worker is available performance is no worse than BO using expected improvement.