Saved in:
Bibliographic Details
Main Authors: Li, Andy, Durrant, Aiden, Markovic, Milan, Huang, Tianjin, Kundu, Souvik, Chen, Tianlong, Yin, Lu, Leontidis, Georgios
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.13545
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918224950460416
author Li, Andy
Durrant, Aiden
Markovic, Milan
Huang, Tianjin
Kundu, Souvik
Chen, Tianlong
Yin, Lu
Leontidis, Georgios
author_facet Li, Andy
Durrant, Aiden
Markovic, Milan
Huang, Tianjin
Kundu, Souvik
Chen, Tianlong
Yin, Lu
Leontidis, Georgios
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2411_13545
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Pushing the Limits of Sparsity: A Bag of Tricks for Extreme Pruning
Li, Andy
Durrant, Aiden
Markovic, Milan
Huang, Tianjin
Kundu, Souvik
Chen, Tianlong
Yin, Lu
Leontidis, Georgios
Computer Vision and Pattern Recognition
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.
title Pushing the Limits of Sparsity: A Bag of Tricks for Extreme Pruning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2411.13545