Work

Size-independent sample complexity of neural networks

Year: 2019
Type: article
Abstract: We study the sample complexity of learning neural networks by providing new bounds on their Rademacher complexity, assuming norm constraints on the parameter matrix of each layer. Compared to previous work, these complexity bounds have improved dependence on the network depth and, under some additional assumptions, are fully independent of the network size (both depth and width). These results are derived using some novel techniques, which may be of independent interest.
Language: English
Open Access status: green
OpenAlex ID: W2777138330
FWCI: 22.17
Cites: 35
Cited by: 412
Related to: 10