Minimal Effort Back Propagation for Convolutional Neural Networks

TL;DR

This paper extends a gradient-sampling technique to CNNs, showing that passing back only 5% of gradients can achieve comparable or better performance, significantly reducing computational costs.

Contribution

The paper introduces a minimal effort backpropagation method for CNNs, demonstrating effective gradient reduction without sacrificing accuracy.

Findings

Passing back 5% of gradients maintains model performance

Gradient top-k selection induces sparse calculations

Significant computational savings in CNN backpropagation

Abstract

As traditional neural network consumes a significant amount of computing resources during back propagation, \citet{Sun2017mePropSB} propose a simple yet effective technique to alleviate this problem. In this technique, only a small subset of the full gradients are computed to update the model parameters. In this paper we extend this technique into the Convolutional Neural Network(CNN) to reduce calculation in back propagation, and the surprising results verify its validity in CNN: only 5\% of the gradients are passed back but the model still achieves the same effect as the traditional CNN, or even better. We also show that the top-$k$ selection of gradients leads to a sparse calculation in back propagation, which may bring significant computational benefits for high computational complexity of convolution operation in CNN.