On the Flatness of Loss Surface for Two-layered ReLU Networks

Jiezhang Cao, Qingyao Wu, Yuguang Yan, Li Wang, Mingkui Tan

Abstract

Deep learning has achieved unprecedented practical success in many applications. Despite its empirical success, however, the theoretical understanding of deep neural networks still remains a major ...
BibTeX
@inproceedings{cao2017flatness,
  author = {Jiezhang Cao and Qingyao Wu and Yuguang Yan and Li Wang and Mingkui Tan},
  title = {{On the Flatness of Loss Surface for Two-layered ReLU Networks}},
  booktitle = {ACML},
  year = {2017},
}
Figure