Learning Sparse Neural Networks via Sensitivity-Driven Regularization

TL;DR

This paper introduces a sensitivity-driven regularization method for training sparse neural networks, effectively reducing parameters while maintaining accuracy, and outperforming recent sparsification techniques in both sparsity and error rates.

Contribution

The paper proposes a novel regularization approach based on output sensitivity to selectively prune neural network weights, achieving higher sparsity with comparable or better accuracy.

Findings

Method surpasses recent techniques in sparsity and error rates.

Achieves up to twice the sparsity at the same error rate.

Effectively prunes parameters with minimal impact on performance.

Abstract

The ever-increasing number of parameters in deep neural networks poses challenges for memory-limited applications. Regularize-and-prune methods aim at meeting these challenges by sparsifying the network weights. In this context we quantify the output sensitivity to the parameters (i.e. their relevance to the network output) and introduce a regularization term that gradually lowers the absolute value of parameters with low sensitivity. Thus, a very large fraction of the parameters approach zero and are eventually set to zero by simple thresholding. Our method surpasses most of the recent techniques both in terms of sparsity and error rates. In some cases, the method reaches twice the sparsity obtained by other techniques at equal error rates.