Sacrificial Learning in Nonlinear Perceptrons

Thumbnail Image

Date

Authors

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Description

Using the cavity method we consider the learning of noisy teacher-generated examples by a nonlinear student perceptron. For insufficient examples and weak weight decay, the activation distribution of the training examples exhibits a gap for the more difficult examples. This illustrates that the outliers are sacrificed for the overall performance. Simulation shows that the picture of the smooth energy landscape cannot describe the gapped distributions well, implying that a rough energy landscape may complicate the learning process.
7 pages, 4 figures

Keywords

Disordered Systems and Neural Networks

Citation

URI

http://salesiana.dossiersoluciones.com/handle/123456789/16472

Consulte el texto completo en el siguiente enlace:

https://arxiv.org/abs/cond-mat/0006206
 http://arxiv.org/abs/cond-mat/0006206

Collections

arXiv