Neural Networks with Quadratic VC Dimension
Abstract
This paper shows that neural networks which use continuous acti(cid:173) vation functions have VC dimension at least as large as the square of the number of weights w. This result settles a long-standing open question, namely whether the well-known O( w log w) bound, known for hard-threshold nets, also held for more general sigmoidal nets. Implications for the number of samples needed for valid gen(cid:173) eralization are discussed.
Cite
Text
Koiran and Sontag. "Neural Networks with Quadratic VC Dimension." Neural Information Processing Systems, 1995.Markdown
[Koiran and Sontag. "Neural Networks with Quadratic VC Dimension." Neural Information Processing Systems, 1995.](https://mlanthology.org/neurips/1995/koiran1995neurips-neural/)BibTeX
@inproceedings{koiran1995neurips-neural,
title = {{Neural Networks with Quadratic VC Dimension}},
author = {Koiran, Pascal and Sontag, Eduardo D.},
booktitle = {Neural Information Processing Systems},
year = {1995},
pages = {197-203},
url = {https://mlanthology.org/neurips/1995/koiran1995neurips-neural/}
}