A Fast Stochastic Error-Descent Algorithm for Supervised Learning and Optimization

NeurIPS 1992 pp. 244-251

/neurips/1992/cauwenberghs1992neurips-fast/

Abstract

A parallel stochastic algorithm is investigated for error-descent learning and optimization in deterministic networks of arbitrary topology. No explicit information about internal network struc(cid:173) ture is needed. The method is based on the model-free distributed learning mechanism of Dembo and Kailath. A modified parameter update rule is proposed by which each individual parameter vector perturbation contributes a decrease in error. A substantially faster learning speed is hence allowed. Furthermore, the modified algo(cid:173) rithm supports learning time-varying features in dynamical net(cid:173) works. We analyze the convergence and scaling properties of the algorithm, and present simulation results for dynamic trajectory learning in recurrent networks.

PDF NeurIPS Semantic Scholar

Cite

Text

Cauwenberghs. "A Fast Stochastic Error-Descent Algorithm for Supervised Learning and Optimization." Neural Information Processing Systems, 1992.

Markdown

[Cauwenberghs. "A Fast Stochastic Error-Descent Algorithm for Supervised Learning and Optimization." Neural Information Processing Systems, 1992.](https://mlanthology.org/neurips/1992/cauwenberghs1992neurips-fast/)

BibTeX

@inproceedings{cauwenberghs1992neurips-fast,
  title     = {{A Fast Stochastic Error-Descent Algorithm for Supervised Learning and Optimization}},
  author    = {Cauwenberghs, Gert},
  booktitle = {Neural Information Processing Systems},
  year      = {1992},
  pages     = {244-251},
  url       = {https://mlanthology.org/neurips/1992/cauwenberghs1992neurips-fast/}
}