Deep Learning with Limited Numerical Precision

Training of large-scale deep neural networks is often constrained by the available computational resources. We study the effect of limited precision data representation and computation on neural network training. Within the context of low-precision fixed-point computations, we observe the rounding s...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2015-02
Main Authors: Gupta, Suyog, Agrawal, Ankur, Gopalakrishnan, Kailash, Narayanan, Pritish
Format: Article
Language:English
Subjects:
Deep learning
Fixed point arithmetic
Neural networks
Representations
Rounding
Training
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Training of large-scale deep neural networks is often constrained by the available computational resources. We study the effect of limited precision data representation and computation on neural network training. Within the context of low-precision fixed-point computations, we observe the rounding scheme to play a crucial role in determining the network's behavior during training. Our results show that deep networks can be trained using only 16-bit wide fixed-point number representation when using stochastic rounding, and incur little to no degradation in the classification accuracy. We also demonstrate an energy-efficient hardware accelerator that implements low-precision fixed-point arithmetic with stochastic rounding.
ISSN:2331-8422