HYPPO: A Surrogate-Based Multi-Level Parallelism Tool for Hyperparameter Optimization
This provides a more efficient hyperparameter tuning tool for deep learning practitioners, though it appears incremental as it builds on existing HPO methods with new parallelism and surrogate techniques.
The paper tackles hyperparameter optimization for deep learning models by introducing HYPPO, a tool that uses adaptive surrogate models and uncertainty quantification to reduce the number of evaluations by an order of magnitude and throughput by two orders of magnitude.
We present a new software, HYPPO, that enables the automatic tuning of hyperparameters of various deep learning (DL) models. Unlike other hyperparameter optimization (HPO) methods, HYPPO uses adaptive surrogate models and directly accounts for uncertainty in model predictions to find accurate and reliable models that make robust predictions. Using asynchronous nested parallelism, we are able to significantly alleviate the computational burden of training complex architectures and quantifying the uncertainty. HYPPO is implemented in Python and can be used with both TensorFlow and PyTorch libraries. We demonstrate various software features on time-series prediction and image classification problems as well as a scientific application in computed tomography image reconstruction. Finally, we show that (1) we can reduce by an order of magnitude the number of evaluations necessary to find the most optimal region in the hyperparameter space and (2) we can reduce by two orders of magnitude the throughput for such HPO process to complete.