Memory-Optimized Once-For-All Network

TL;DR

This paper introduces Memory-Optimized OFA (MOOFA), a supernet that better utilizes device memory to improve DNN deployment efficiency and accuracy on resource-constrained hardware.

Contribution

We propose MOOFA, a novel supernet that maximizes memory usage for improved model generalizability and performance on limited-resource devices.

Findings

MOOFA outperforms original OFA in memory utilization.

MOOFA achieves higher accuracy on ImageNet.

Enhanced features diversity in MOOFA improves deployment efficiency.

Abstract

Deploying Deep Neural Networks (DNNs) on different hardware platforms is challenging due to varying resource constraints. Besides handcrafted approaches aiming at making deep models hardware-friendly, Neural Architectures Search is rising as a toolbox to craft more efficient DNNs without sacrificing performance. Among these, the Once-For-All (OFA) approach offers a solution by allowing the sampling of well-performing sub-networks from a single supernet -- this leads to evident advantages in terms of computation. However, OFA does not fully utilize the potential memory capacity of the target device, focusing instead on limiting maximum memory usage per layer. This leaves room for an unexploited potential in terms of model generalizability. In this paper, we introduce a Memory-Optimized OFA (MOOFA) supernet, designed to enhance DNN deployment on resource-limited devices by maximizing memory usage (and for instance, features diversity) across different configurations. Tested on ImageNet, our MOOFA supernet demonstrates improvements in memory exploitation and model accuracy compared to the original OFA supernet. Our code is available at https://github.com/MaximeGirard/memory-optimized-once-for-all.