ML Anthology

Improving Generalization Capabilities of Dynamic Neural Networks

Miroslaw Galicki, Lutz Leistritz, Ernst Bernhard Zwick, Herbert Witte
NeCo 2004 pp. 1253-1282
doi:10.1162/089976604773717603 /neco/2004/galicki2004neco-improving/

Abstract

This work addresses the problem of improving the generalization capabilities of continuous recurrent neural networks. The learning task is transformed into an optimal control framework in which the weights and the initial network state are treated as unknown controls. A new learning algorithm based on a variational formulation of Pontrayagin's maximum principle is proposed. Under reasonable assumptions, its convergence is discussed. Numerical examples are given that demonstrate an essential improvement of generalization capabilities after the learning process of a dynamic network.

NeCo Semantic Scholar

Cite

Text

Galicki et al. "Improving Generalization Capabilities of Dynamic Neural Networks." Neural Computation, 2004. doi:10.1162/089976604773717603

Markdown

[Galicki et al. "Improving Generalization Capabilities of Dynamic Neural Networks." Neural Computation, 2004.](https://mlanthology.org/neco/2004/galicki2004neco-improving/) doi:10.1162/089976604773717603

BibTeX

@article{galicki2004neco-improving,
  title     = {{Improving Generalization Capabilities of Dynamic Neural Networks}},
  author    = {Galicki, Miroslaw and Leistritz, Lutz and Zwick, Ernst Bernhard and Witte, Herbert},
  journal   = {Neural Computation},
  year      = {2004},
  pages     = {1253-1282},
  doi       = {10.1162/089976604773717603},
  volume    = {16},
  url       = {https://mlanthology.org/neco/2004/galicki2004neco-improving/}
}