Implementation of boolean neural networks on parallel computers

This paper analyses the parallel implementation using networks of transputers of a neural structure belonging to a particular class of neural architectures known as GSN neural networks. These architectures, belonging to the general clasa of RAM-based networks and composed 01 digitally specified proc...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Carvalho, Andre Carlos Ponce de Leon Ferreira de, Fairhurst, Michael C., Bisset, David L.
Formato: Objeto de conferencia
Lenguaje:Inglés
Publicado: 1996
Materias:
Acceso en línea:http://sedici.unlp.edu.ar/handle/10915/24161
Aporte de:
Descripción
Sumario:This paper analyses the parallel implementation using networks of transputers of a neural structure belonging to a particular class of neural architectures known as GSN neural networks. These architectures, belonging to the general clasa of RAM-based networks and composed 01 digitally specified processing nodes, have been implemented using different processing topologies, and performance in relatíon to both training and testing efficiency in a practical pattern recognition task has been evaluated.