Multi-Level Residual Networks from Dynamical Systems View
Abstract
Deep residual networks (ResNets) and their variants are widely used in many computer vision applications and natural language processing tasks. However, the theoretical principles for designing and training ResNets are still not fully understood. Recently, several points of view have emerged to try to interpret ResNet theoretically, such as unraveled view, unrolled iterative estimation and dynamical systems view. In this paper, we adopt the dynamical systems point of view, and analyze the lesioning properties of ResNet both theoretically and experimentally. Based on these analyses, we additionally propose a novel method for accelerating ResNet training. We apply the proposed method to train ResNets and Wide ResNets for three image classification benchmarks, reducing training time by more than 40\% with superior or on-par accuracy.
Cite
Text
Chang et al. "Multi-Level Residual Networks from Dynamical Systems View." International Conference on Learning Representations, 2018.Markdown
[Chang et al. "Multi-Level Residual Networks from Dynamical Systems View." International Conference on Learning Representations, 2018.](https://mlanthology.org/iclr/2018/chang2018iclr-multilevel/)BibTeX
@inproceedings{chang2018iclr-multilevel,
title = {{Multi-Level Residual Networks from Dynamical Systems View}},
author = {Chang, Bo and Meng, Lili and Haber, Eldad and Tung, Frederick and Begert, David},
booktitle = {International Conference on Learning Representations},
year = {2018},
url = {https://mlanthology.org/iclr/2018/chang2018iclr-multilevel/}
}