Probing the basins of attraction of a recurrent neural network

Authors
Publication date 2000
Journal Journal of Physics. A, Mathematical and General
Volume | Issue number 33
Pages (from-to) 1765-1780
Organisations
  • Faculty of Science (FNWI) - Institute of Physics (IoP) - Institute for Theoretical Physics Amsterdam (ITFA)
Abstract Analytical expressions for the weights $w_{ij}(b)$ of the connections of a recurrent neural network are found by taking explicitly into account basins of attraction, the size of which is characterized by a basin parameter $b$. It is shown that a network with $b \neq 0$ performs better than one with $b=0$: a network that has to retrieve perturbed patterns must
be constructed with ---similarly--- perturbed patterns.
Document type Article
Published at https://doi.org/10.1088/0305-4470/33/9/304
Permalink to this page
Back