EXPObench: Benchmarking Surrogate-based Optimisation Algorithms on Expensive Black-box Functions
This work addresses a standardization problem for researchers and practitioners in optimization, offering a benchmark and dataset to facilitate more uniform analysis and reduce the need for expensive evaluations, though it is incremental in nature.
The authors tackled the lack of standardization in benchmarking surrogate-based optimization algorithms for expensive black-box functions by introducing EXPObench, a new benchmark library that includes real-life applications, and used it to compare six algorithms, providing new insights and rules of thumb for algorithm selection.
Surrogate algorithms such as Bayesian optimisation are especially designed for black-box optimisation problems with expensive objectives, such as hyperparameter tuning or simulation-based optimisation. In the literature, these algorithms are usually evaluated with synthetic benchmarks which are well established but have no expensive objective, and only on one or two real-life applications which vary wildly between papers. There is a clear lack of standardisation when it comes to benchmarking surrogate algorithms on real-life, expensive, black-box objective functions. This makes it very difficult to draw conclusions on the effect of algorithmic contributions and to give substantial advice on which method to use when. A new benchmark library, EXPObench, provides first steps towards such a standardisation. The library is used to provide an extensive comparison of six different surrogate algorithms on four expensive optimisation problems from different real-life applications. This has led to new insights regarding the relative importance of exploration, the evaluation time of the objective, and the used model. We also provide rules of thumb for which surrogate algorithm to use in which situation. A further contribution is that we make the algorithms and benchmark problem instances publicly available, contributing to more uniform analysis of surrogate algorithms. Most importantly, we include the performance of the six algorithms on all evaluated problem instances. This results in a unique new dataset that lowers the bar for researching new methods as the number of expensive evaluations required for comparison is significantly reduced.