Structured Network Pruning by Measuring Filter-wise Interactions

TL;DR

This paper introduces SNPFI, a structured network pruning method that measures filter interactions to more effectively identify redundant filters, reducing computation by about 60% while maintaining accuracy.

Contribution

It proposes a novel redundancy criterion incorporating filter interactions and a pruning approach that adaptively assigns sparsity and recovers performance without iterative training.

Findings

Achieves nearly 60% computation reduction across multiple CNN models.

Maintains classification accuracy after pruning.

Effective on datasets like MNIST, CIFAR-10, and ImageNet.

Abstract

Structured network pruning is a practical approach to reduce computation cost directly while retaining the CNNs' generalization performance in real applications. However, identifying redundant filters is a core problem in structured network pruning, and current redundancy criteria only focus on individual filters' attributes. When pruning sparsity increases, these redundancy criteria are not effective or efficient enough. Since the filter-wise interaction also contributes to the CNN's prediction accuracy, we integrate the filter-wise interaction into the redundancy criterion. In our criterion, we introduce the filter importance and filter utilization strength to reflect the decision ability of individual and multiple filters. Utilizing this new redundancy criterion, we propose a structured network pruning approach SNPFI (Structured Network Pruning by measuring Filter-wise Interaction). During the pruning, the SNPFI can automatically assign the proper sparsity based on the filter utilization strength and eliminate the useless filters by filter importance. After the pruning, the SNPFI can recover pruned model's performance effectively without iterative training by minimizing the interaction difference. We empirically demonstrate the effectiveness of the SNPFI with several commonly used CNN models, including AlexNet, MobileNetv1, and ResNet-50, on various image classification datasets, including MNIST, CIFAR-10, and ImageNet. For all experimental CNN models, nearly 60% of computation is reduced in a network compression while the classification accuracy remains.