Scaling Deep Networks with the Mesh Adaptive Direct Search algorithm
This work addresses the challenge of deploying large neural networks on edge and IoT devices by providing an incremental method for network compression.
The paper tackles the problem of designing lightweight deep neural networks for image classification by using the Mesh Adaptive Direct Search (MADS) algorithm to automate the design process, resulting in competitive compression rates with fewer trials.
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.