PriorBand: Practical Hyperparameter Optimization in the Age of Deep Learning

Neeratyoy Mallik, Eddie Bergman, Carl Hvarfner, Danny Stoll, Maciej Janowski, Marius Lindauer, Luigi Nardi, Frank Hutter

Keywords: Hyperparameter Optimization, Deep Learning

TL;DR: We propose a hyperparameter optimization algorithm tailored to the needs of deep learning researchers & practitioners that draws on expert beliefs and cheap preliminary explorations, and show its robustness and efficiency.

Abstract:

Hyperparameters of Deep Learning (DL) pipelines are crucial for their downstream performance.
While a large number of methods for Hyperparameter Optimization (HPO) have been developed, their incurred costs are often untenable for modern DL.
Consequently, manual experimentation is still the most prevalent approach to optimize hyperparameters, relying on the researcher's intuition, domain knowledge, and cheap preliminary explorations.
To resolve this misalignment between HPO algorithms and DL researchers, we propose PriorBand, an HPO algorithm tailored to DL, able to utilize both expert beliefs and cheap proxy tasks.
Empirically, we demonstrate PriorBand's efficiency across a range of DL benchmarks and show its gains under informative expert input and robustness against poor expert beliefs.

Submission Number: 10593