Explaining Landscape Connectivity of Low-Cost Solutions for Multilayer Nets
Abstract
Mode connectivity is a surprising phenomenon in the loss landscape of deep nets. Optima---at least those discovered by gradient-based optimization---turn out to be connected by simple paths on which the loss function is almost constant. Often, these paths can be chosen to be piece-wise linear, with as few as two segments. We give mathematical explanations for this phenomenon, assuming generic properties (such as dropout stability and noise stability) of well-trained deep nets, which have previously been identified as part of understanding the generalization properties of deep nets. Our explanation holds for realistic multilayer nets, and experiments are presented to verify the theory.
Cite
Text
Kuditipudi et al. "Explaining Landscape Connectivity of Low-Cost Solutions for Multilayer Nets." Neural Information Processing Systems, 2019.Markdown
[Kuditipudi et al. "Explaining Landscape Connectivity of Low-Cost Solutions for Multilayer Nets." Neural Information Processing Systems, 2019.](https://mlanthology.org/neurips/2019/kuditipudi2019neurips-explaining/)BibTeX
@inproceedings{kuditipudi2019neurips-explaining,
title = {{Explaining Landscape Connectivity of Low-Cost Solutions for Multilayer Nets}},
author = {Kuditipudi, Rohith and Wang, Xiang and Lee, Holden and Zhang, Yi and Li, Zhiyuan and Hu, Wei and Ge, Rong and Arora, Sanjeev},
booktitle = {Neural Information Processing Systems},
year = {2019},
pages = {14601-14610},
url = {https://mlanthology.org/neurips/2019/kuditipudi2019neurips-explaining/}
}