Bayesian Optimization is Superior to Random Search for Machine Learning Hyperparameter Tuning: Analysis of the Black-Box Optimization Challenge 2020

TL;DR

This paper analyzes the results of the NeurIPS 2020 black-box optimization challenge, demonstrating that Bayesian optimization outperforms random search in hyperparameter tuning for machine learning models.

Contribution

It provides the first comprehensive comparison of black-box optimization methods, highlighting the superiority of Bayesian optimization for hyperparameter tuning in real-world ML tasks.

Findings

Bayesian optimization outperforms random search in hyperparameter tuning

Black-box optimization methods are effective for real-world ML model tuning

The challenge results inform best practices for hyperparameter optimization

Abstract

This paper presents the results and insights from the black-box optimization (BBO) challenge at NeurIPS 2020 which ran from July-October, 2020. The challenge emphasized the importance of evaluating derivative-free optimizers for tuning the hyperparameters of machine learning models. This was the first black-box optimization challenge with a machine learning emphasis. It was based on tuning (validation set) performance of standard machine learning models on real datasets. This competition has widespread impact as black-box optimization (e.g., Bayesian optimization) is relevant for hyperparameter tuning in almost every machine learning project as well as many applications outside of machine learning. The final leaderboard was determined using the optimization performance on held-out (hidden) objective functions, where the optimizers ran without human intervention. Baselines were set using the default settings of several open-source black-box optimization packages as well as random search.