ChannelNets: Compact and Efficient Convolutional Neural Networks via Channel-Wise Convolutions

Abstract

Convolutional neural networks (CNNs) have shown great capability of solving various artificial intelligence tasks. However, the increasing model size has raised challenges in employing them in resource-limited applications. In this work, we propose to compress deep models by using channel-wise convolutions, which re- place dense connections among feature maps with sparse ones in CNNs. Based on this novel operation, we build light-weight CNNs known as ChannelNets. Channel- Nets use three instances of channel-wise convolutions; namely group channel-wise convolutions, depth-wise separable channel-wise convolutions, and the convolu- tional classification layer. Compared to prior CNNs designed for mobile devices, ChannelNets achieve a significant reduction in terms of the number of parameters and computational cost without loss in accuracy. Notably, our work represents the first attempt to compress the fully-connected classification layer, which usually accounts for about 25% of total parameters in compact CNNs. Experimental results on the ImageNet dataset demonstrate that ChannelNets achieve consistently better performance compared to prior methods.

arXiv comment10 pages, NIPS18

Cited by in corpus (10)

• ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks
• EDropout: Energy-Based Dropout and Pruning of Deep Neural Networks
• Kronecker Attention Networks
• Learning Versatile Convolution Filters for Efficient Visual Recognition
• Neural Epitome Search for Architecture-Agnostic Network Compression
• NAS-DIP: Learning Deep Image Prior with Neural Architecture Search
• Separable Layers Enable Structured Efficient Linear Substitutions
• Cross-Channel Intragroup Sparsity Neural Network
• Full-Stack Filters to Build Minimum Viable CNNs
• Self-grouping Convolutional Neural Networks