E2-Train: Training State-of-the-art CNNs with Over 80% Energy Savings

TL;DR

This paper introduces methods to significantly reduce energy consumption during CNN training, enabling more sustainable on-device training with minimal accuracy loss, validated through extensive simulations and real FPGA measurements.

Contribution

It proposes a novel multi-level approach to decrease training energy, including data, model, and algorithm optimizations, demonstrating substantial energy savings with minimal accuracy impact.

Findings

Over 90% energy savings for ResNet-74 on CIFAR-10.

Over 84% energy savings for ResNet-110 on CIFAR-100.

Minimal accuracy loss (<2%) with significant energy reduction.

Abstract

Convolutional neural networks (CNNs) have been increasingly deployed to edge devices. Hence, many efforts have been made towards efficient CNN inference in resource-constrained platforms. This paper attempts to explore an orthogonal direction: how to conduct more energy-efficient training of CNNs, so as to enable on-device training. We strive to reduce the energy cost during training, by dropping unnecessary computations from three complementary levels: stochastic mini-batch dropping on the data level; selective layer update on the model level; and sign prediction for low-cost, low-precision back-propagation, on the algorithm level. Extensive simulations and ablation studies, with real energy measurements from an FPGA board, confirm the superiority of our proposed strategies and demonstrate remarkable energy savings for training. For example, when training ResNet-74 on CIFAR-10, we achieve aggressive energy savings of >90% and >60%, while incurring a top-1 accuracy loss of only about 2% and 1.2%, respectively. When training ResNet-110 on CIFAR-100, an over 84% training energy saving is achieved without degrading inference accuracy.