CHIP: CHannel Independence-based Pruning for Compact Neural Networks

TL;DR

This paper introduces a novel filter pruning method based on channel independence, which measures correlations among feature maps to identify less useful filters, leading to more compact neural networks with maintained or improved accuracy.

Contribution

The paper proposes a new inter-channel based filter pruning approach using channel independence, demonstrating superior performance over existing methods on multiple datasets.

Findings

Achieves up to 0.94% accuracy increase on CIFAR-10 models.

Reduces model size and FLOPs by over 42% on CIFAR-10.

Reduces storage and computation by over 40% on ImageNet with minimal accuracy loss.

Abstract

Filter pruning has been widely used for neural network compression because of its enabled practical acceleration. To date, most of the existing filter pruning works explore the importance of filters via using intra-channel information. In this paper, starting from an inter-channel perspective, we propose to perform efficient filter pruning using Channel Independence, a metric that measures the correlations among different feature maps. The less independent feature map is interpreted as containing less useful information$/$knowledge, and hence its corresponding filter can be pruned without affecting model capacity. We systematically investigate the quantification metric, measuring scheme and sensitiveness$/$reliability of channel independence in the context of filter pruning. Our evaluation results for different models on various datasets show the superior performance of our approach. Notably, on CIFAR-10 dataset our solution can bring $0.90\%$ and $0.94\%$ accuracy increase over baseline ResNet-56 and ResNet-110 models, respectively, and meanwhile the model size and FLOPs are reduced by $42.8\%$ and $47.4\%$ (for ResNet-56) and $48.3\%$ and $52.1\%$ (for ResNet-110), respectively. On ImageNet dataset, our approach can achieve $40.8\%$ and $44.8\%$ storage and computation reductions, respectively, with $0.15\%$ accuracy increase over the baseline ResNet-50 model. The code is available at https://github.com/Eclipsess/CHIP_NeurIPS2021.