Dynamic ReLU
Abstract
Rectified linear units (ReLU) are commonly used in deep neural networks. So far ReLU and its generalizations (non-parametric or parametric) are static, performing identically for all input samples. In this paper, we propose dynamic ReLU (DY-ReLU), a dynamic rectifier of which parameters are generated by a hyper function over all in-put elements. The key insight is that DY-ReLU encodes the global context into the hyper function, and adapts the piecewise linear activation function accordingly. Compared to its static counterpart, DY-ReLU has negligible extra computational cost, but significantly more representation capability, especially for light-weight neural networks. By simply using DY-ReLU for MobileNetV2, the top-1 accuracy on ImageNet classification is boosted from 72.0% to 76.2% with only 5% additional FLOPs.
Keywords
Cite
@article{arxiv.2003.10027,
title = {Dynamic ReLU},
author = {Yinpeng Chen and Xiyang Dai and Mengchen Liu and Dongdong Chen and Lu Yuan and Zicheng Liu},
journal= {arXiv preprint arXiv:2003.10027},
year = {2020}
}
Comments
ECCV 2020