ABCP: Automatic Block-wise and Channel-wise Network Pruning via Joint Search

TL;DR

ABCP uses deep reinforcement learning to automatically perform block-wise and channel-wise network pruning, achieving high compression ratios with minimal accuracy loss, reducing manual effort and improving robustness.

Contribution

It introduces a joint search method for block-wise and channel-wise pruning using reinforcement learning, outperforming traditional rule-based approaches.

Findings

Achieves 99.5% FLOPs reduction on YOLOv3

Reduces parameters by 99.5% with 2.8% mAP loss

Enhances robustness in transfer detection tasks

Abstract

Currently, an increasing number of model pruning methods are proposed to resolve the contradictions between the computer powers required by the deep learning models and the resource-constrained devices. However, most of the traditional rule-based network pruning methods can not reach a sufficient compression ratio with low accuracy loss and are time-consuming as well as laborious. In this paper, we propose Automatic Block-wise and Channel-wise Network Pruning (ABCP) to jointly search the block-wise and channel-wise pruning action with deep reinforcement learning. A joint sample algorithm is proposed to simultaneously generate the pruning choice of each residual block and the channel pruning ratio of each convolutional layer from the discrete and continuous search space respectively. The best pruning action taking both the accuracy and the complexity of the model into account is obtained finally. Compared with the traditional rule-based pruning method, this pipeline saves human labor and achieves a higher compression ratio with lower accuracy loss. Tested on the mobile robot detection dataset, the pruned YOLOv3 model saves 99.5% FLOPs, reduces 99.5% parameters, and achieves 37.3 times speed up with only 2.8% mAP loss. The results of the transfer task on the sim2real detection dataset also show that our pruned model has much better robustness performance.