PriorBand: Practical Hyperparameter Optimization in the Age of Deep Learning
This work addresses the misalignment between existing HPO methods and deep learning researchers, offering a practical solution for optimizing hyperparameters in modern DL pipelines.
The paper tackles the problem of hyperparameter optimization in deep learning by proposing PriorBand, an algorithm that incorporates expert beliefs and cheap proxy tasks, achieving significant efficiency gains across benchmarks.
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