LMEMs for post-hoc analysis of HPO Benchmarking

TL;DR

This paper introduces the use of Linear Mixed-Effect Models for post-hoc analysis of hyperparameter optimization benchmarks, providing more detailed insights than traditional averaging methods.

Contribution

It applies LMEMs to HPO benchmarking data, enabling nuanced analysis that accounts for dataset and method variability, which is a novel approach in this context.

Findings

LMEMs reveal significant differences between HPO methods.

Deeper insights into dataset-specific performance variations.

Enhanced understanding of hyperparameter tuning effectiveness.

Abstract

The importance of tuning hyperparameters in Machine Learning (ML) and Deep Learning (DL) is established through empirical research and applications, evident from the increase in new hyperparameter optimization (HPO) algorithms and benchmarks steadily added by the community. However, current benchmarking practices using averaged performance across many datasets may obscure key differences between HPO methods, especially for pairwise comparisons. In this work, we apply Linear Mixed-Effect Models-based (LMEMs) significance testing for post-hoc analysis of HPO benchmarking runs. LMEMs allow flexible and expressive modeling on the entire experiment data, including information such as benchmark meta-features, offering deeper insights than current analysis practices. We demonstrate this through a case study on the PriorBand paper's experiment data to find insights not reported in the original work.