Pushing the Limits of Sparsity: A Bag of Tricks for Extreme Pruning

TL;DR

This paper introduces novel techniques for training neural networks at extreme sparsity levels (up to 99.99%), enabling models to maintain high accuracy through dynamic training strategies and parameter sharing.

Contribution

The authors develop and validate three new methods—Dynamic ReLU phasing, weight sharing, and cyclic sparsity—that enable stable training of highly sparse neural networks.

Findings

Achieved 99.99% sparsity with maintained accuracy on ResNet architectures.

Demonstrated improved performance over existing methods at extreme sparsity levels.

Validated techniques on CIFAR-10, CIFAR-100, and ImageNet datasets.

Abstract

Pruning of deep neural networks has been an effective technique for reducing model size while preserving most of the performance of dense networks, crucial for deploying models on memory and power-constrained devices. While recent sparse learning methods have shown promising performance up to moderate sparsity levels such as 95% and 98%, accuracy quickly deteriorates when pushing sparsities to extreme levels due to unique challenges such as fragile gradient flow. In this work, we explore network performance beyond the commonly studied sparsities, and develop techniques that encourage stable training without accuracy collapse even at extreme sparsities, including 99.90%, 99.95\% and 99.99% on ResNet architectures. We propose three complementary techniques that enhance sparse training through different mechanisms: 1) Dynamic ReLU phasing, where DyReLU initially allows for richer parameter exploration before being gradually replaced by standard ReLU, 2) weight sharing which reuses parameters within a residual layer while maintaining the same number of learnable parameters, and 3) cyclic sparsity, where both sparsity levels and sparsity patterns evolve dynamically throughout training to better encourage parameter exploration. We evaluate our method, which we term Extreme Adaptive Sparse Training (EAST) at extreme sparsities using ResNet-34 and ResNet-50 on CIFAR-10, CIFAR-100, and ImageNet, achieving competitive or improved performance compared to existing methods, with notable gains at extreme sparsity levels.