Group Pruning using a Bounded-Lp norm for Group Gating and Regularization

TL;DR

This paper introduces a group-wise pruning method using a bounded-Lp norm for gating and regularization, significantly reducing model size while maintaining accuracy across multiple neural network architectures and datasets.

Contribution

It proposes a novel bounded-Lp regularizer and gating mechanism for effective group pruning during training, improving network efficiency without sacrificing performance.

Findings

ResNet-164 parameters reduced by 30% on CIFAR100

DenseNet-40 parameters reduced by 69% on CIFAR100

MobileNetV2 parameters reduced by 75% on CIFAR100 and further compressed on ImageNet

Abstract

Deep neural networks achieve state-of-the-art results on several tasks while increasing in complexity. It has been shown that neural networks can be pruned during training by imposing sparsity inducing regularizers. In this paper, we investigate two techniques for group-wise pruning during training in order to improve network efficiency. We propose a gating factor after every convolutional layer to induce channel level sparsity, encouraging insignificant channels to become exactly zero. Further, we introduce and analyse a bounded variant of the L1 regularizer, which interpolates between L1 and L0-norms to retain performance of the network at higher pruning rates. To underline effectiveness of the proposed methods,we show that the number of parameters of ResNet-164, DenseNet-40 and MobileNetV2 can be reduced down by 30%, 69% and 75% on CIFAR100 respectively without a significant drop in accuracy. We achieve state-of-the-art pruning results for ResNet-50 with higher accuracy on ImageNet. Furthermore, we show that the light weight MobileNetV2 can further be compressed on ImageNet without a significant drop in performance.