Reduced-Precision Strategies for Bounded Memory in Deep Neural Nets

TL;DR

This paper explores layer-specific reduced-precision data strategies in CNNs, demonstrating significant data footprint reductions with minimal accuracy loss by tuning precision per layer.

Contribution

It introduces a method for optimizing layer-wise precision in CNNs, enabling substantial data footprint reduction while maintaining high accuracy.

Findings

Average 74% reduction in data footprint with less than 1% accuracy loss

Layer-wise error tolerance varies within networks

Proposed method effectively finds low precision configurations

Abstract

This work investigates how using reduced precision data in Convolutional Neural Networks (CNNs) affects network accuracy during classification. More specifically, this study considers networks where each layer may use different precision data. Our key result is the observation that the tolerance of CNNs to reduced precision data not only varies across networks, a well established observation, but also within networks. Tuning precision per layer is appealing as it could enable energy and performance improvements. In this paper we study how error tolerance across layers varies and propose a method for finding a low precision configuration for a network while maintaining high accuracy. A diverse set of CNNs is analyzed showing that compared to a conventional implementation using a 32-bit floating-point representation for all layers, and with less than 1% loss in relative accuracy, the data footprint required by these networks can be reduced by an average of 74% and up to 92%.