An approach for construction of Boolean neural networks based on geometrical expansion

Di Wang, Narendra S. Chaudhari

Research output: Contribution to journalArticlepeer-review

20 Scopus citations

Abstract

We propose a fast covering learning algorithm (FCLA) for construction of Boolean neural networks. We visualize a neuron in terms of a hypersphere. To expand this hypersphere, we introduce three different radii. The construction process makes use of three concentric hyperspheres based on these radii, and is illustrated using an example. FCLA results in a simple neural network; further the resulting network structure is less sensitive to the order in which the input is given.

Original languageBritish English
Pages (from-to)455-461
Number of pages7
JournalNeurocomputing
Volume57
Issue number1-4
DOIs
StatePublished - Mar 2004

Keywords

  • Binary neural networks
  • Geometric learning
  • Hard limiter neuron
  • Linear separability

Fingerprint

Dive into the research topics of 'An approach for construction of Boolean neural networks based on geometrical expansion'. Together they form a unique fingerprint.

Cite this