A case for efficient accelerator design space exploration via Bayesian optimization
Brandon Reagen, José Miguel Hernández-Lobato, Robert Adolf, Michael A. Gelbart, Paul N. Whatmough, Gu-Yeon Wei, David M. Brooks · 2017
In this paper we propose using machine learning to improve the design of deep neural network hardware accelerators. We show how to adapt multi-objective Bayesian optimization to overcome a challenging design problem: optimizing deep neural network hardware accelerators for both accuracy and energy efficiency. DNN accelerators exhibit all aspects of a challenging optimization space: the landscape is rough, evaluating designs is expensive, the objectives compete with each other, and both design spaces (algorithmic and microarchitectural) are unwieldy. With multi-objective Bayesian optimization, the design space exploration is made tractable and the design points found vastly outperform traditional methods across all metrics of interest.