HongyangGao / ChannelNets

Tensorflow Implementation of ChannelNets (NeurIPS 18)

Home Page:https://papers.nips.cc/paper/7766-channelnets-compact-and-efficient-convolutional-neural-networks-via-channel-wise-convolutions.pdf

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

ChannelNets

Created by Hongyang Gao, Zhengyang Wang, and Shuiwang Ji at Texas A&M University.

Introduction

ChannelNets are compact and efficent CNN via Channel-wise convolutions. It has been accepted in NIPS2018.

Detailed information about ChannelNets is provided in https://papers.nips.cc/paper/7766-channelnets-compact-and-efficient-convolutional-neural-networks-via-channel-wise-convolutions.pdf.

Citation

@inproceedings{gao2018channelnets,
  title={ChannelNets: Compact and Efficient Convolutional Neural Networks via Channel-Wise Convolutions},
  author={Gao, Hongyang and Wang, Zhengyang and Ji, Shuiwang},
  booktitle={Advances in Neural Information Processing Systems},
  pages={5203--5211},
  year={2018}
}

Results

Models Top-1 Params FLOPs
GoogleNet 0.698 6.8m 1550m
VGG16 0.715 128m 15300m
AlexNet 0.572 60m 720m
SqueezeNet 0.575 1.3m 833m
1.0 MobileNet 0.706 4.2m 569m
ShuffleNet 2x 0.709 5.3m 524m
ChannelNet-v1 0.705 3.7m 407m

Configure the network

All network hyperparameters are configured in main.py.