Enabling Binary Neural Network Training on the Edge

TL;DR

This paper introduces a memory-efficient method for training binary neural networks directly on edge devices, enabling on-device learning with minimal accuracy loss and significant memory and energy savings.

Contribution

The authors demonstrate that backward propagation in binary neural networks is robust to quantization and propose a low-cost training strategy that reduces memory usage by 3-5 times without accuracy loss.

Findings

Memory requirement reductions of 3-5× compared to standard methods

Achieved similar accuracy with significantly less memory on small models

Successfully trained binarized ResNet-18 on ImageNet with 3.78× memory reduction

Abstract

The ever-growing computational demands of increasingly complex machine learning models frequently necessitate the use of powerful cloud-based infrastructure for their training. Binary neural networks are known to be promising candidates for on-device inference due to their extreme compute and memory savings over higher-precision alternatives. However, their existing training methods require the concurrent storage of high-precision activations for all layers, generally making learning on memory-constrained devices infeasible. In this article, we demonstrate that the backward propagation operations needed for binary neural network training are strongly robust to quantization, thereby making on-the-edge learning with modern models a practical proposition. We introduce a low-cost binary neural network training strategy exhibiting sizable memory footprint reductions while inducing little to no accuracy loss vs Courbariaux & Bengio's standard approach. These decreases are primarily enabled through the retention of activations exclusively in binary format. Against the latter algorithm, our drop-in replacement sees memory requirement reductions of 3--5$\times$, while reaching similar test accuracy in comparable time, across a range of small-scale models trained to classify popular datasets. We also demonstrate from-scratch ImageNet training of binarized ResNet-18, achieving a 3.78$\times$ memory reduction. Our work is open-source, and includes the Raspberry Pi-targeted prototype we used to verify our modeled memory decreases and capture the associated energy drops. Such savings will allow for unnecessary cloud offloading to be avoided, reducing latency, increasing energy efficiency, and safeguarding end-user privacy.