Scaling Deep Networks with the Mesh Adaptive Direct Search algorithm

TL;DR

This paper introduces the use of the Mesh Adaptive Direct Search algorithm to automate the design of lightweight deep neural networks for image classification, effectively handling expensive evaluations and constraints.

Contribution

It applies the MADS derivative-free optimization method to neural network design, demonstrating competitive results with fewer trials compared to traditional methods.

Findings

Achieved competitive compression rates.

Reduced number of trials needed for design.

Effectively handled expensive blackbox evaluations.

Abstract

Deep neural networks are getting larger. Their implementation on edge and IoT devices becomes more challenging and moved the community to design lighter versions with similar performance. Standard automatic design tools such as \emph{reinforcement learning} and \emph{evolutionary computing} fundamentally rely on cheap evaluations of an objective function. In the neural network design context, this objective is the accuracy after training, which is expensive and time-consuming to evaluate. We automate the design of a light deep neural network for image classification using the \emph{Mesh Adaptive Direct Search}(MADS) algorithm, a mature derivative-free optimization method that effectively accounts for the expensive blackbox nature of the objective function to explore the design space, even in the presence of constraints.Our tests show competitive compression rates with reduced numbers of trials.