Pruning Convolutional Neural Networks for Resource Efficient Inference

TL;DR

This paper introduces a Taylor expansion-based criterion for pruning convolutional neural networks, enabling resource-efficient inference with minimal accuracy loss, especially effective in transfer learning and large-scale datasets.

Contribution

It presents a novel pruning criterion based on Taylor expansion, combined with greedy pruning and fine-tuning, improving efficiency in transfer learning scenarios.

Findings

Superior pruning performance over traditional criteria.

Achieved over 10x reduction in 3D-convolutional filters.

Effective on large-scale ImageNet dataset.

Abstract

We propose a new formulation for pruning convolutional kernels in neural networks to enable efficient inference. We interleave greedy criteria-based pruning with fine-tuning by backpropagation - a computationally efficient procedure that maintains good generalization in the pruned network. We propose a new criterion based on Taylor expansion that approximates the change in the cost function induced by pruning network parameters. We focus on transfer learning, where large pretrained networks are adapted to specialized tasks. The proposed criterion demonstrates superior performance compared to other criteria, e.g. the norm of kernel weights or feature map activation, for pruning large CNNs after adaptation to fine-grained classification tasks (Birds-200 and Flowers-102) relaying only on the first order gradient information. We also show that pruning can lead to more than 10x theoretical (5x practical) reduction in adapted 3D-convolutional filters with a small drop in accuracy in a recurrent gesture classifier. Finally, we show results for the large-scale ImageNet dataset to emphasize the flexibility of our approach.