ML Anthology

Remarks on Interpolation and Recognition Using Neural Nets

Eduardo D. Sontag
NeurIPS 1990 pp. 939-945
/neurips/1990/sontag1990neurips-remarks/

Abstract

We consider different types of single-hidden-Iayer feedforward nets: with or without direct input to output connections, and using either thresh(cid:173) old or sigmoidal activation functions. The main results show that direct connections in threshold nets double the recognition but not the interpo(cid:173) lation power, while using sigmoids rather than thresholds allows (at least) doubling both. Various results are also given on VC dimension and other measures of recognition capabilities.

PDF NeurIPS Semantic Scholar

Cite

Text

Sontag. "Remarks on Interpolation and Recognition Using Neural Nets." Neural Information Processing Systems, 1990.

Markdown

[Sontag. "Remarks on Interpolation and Recognition Using Neural Nets." Neural Information Processing Systems, 1990.](https://mlanthology.org/neurips/1990/sontag1990neurips-remarks/)

BibTeX

@inproceedings{sontag1990neurips-remarks,
  title     = {{Remarks on Interpolation and Recognition Using Neural Nets}},
  author    = {Sontag, Eduardo D.},
  booktitle = {Neural Information Processing Systems},
  year      = {1990},
  pages     = {939-945},
  url       = {https://mlanthology.org/neurips/1990/sontag1990neurips-remarks/}
}