Ontology highlight
ABSTRACT:
SUBMITTER: Mei S
PROVIDER: S-EPMC6099898 | biostudies-other | 2018 Aug
REPOSITORIES: biostudies-other
Mei Song S Montanari Andrea A Nguyen Phan-Minh PM
Proceedings of the National Academy of Sciences of the United States of America 20180727 33
Multilayer neural networks are among the most powerful models in machine learning, yet the fundamental reasons for this success defy mathematical understanding. Learning a neural network requires optimizing a nonconvex high-dimensional objective (risk function), a problem that is usually attacked using stochastic gradient descent (SGD). Does SGD converge to a global optimum of the risk or only to a local optimum? In the former case, does this happen because local minima are absent or because SGD ...[more]