Título: HENNS: Heterogeneous Environment Neural Network Simulator
Autores: Meira Junior, Wagner; Senna, André Luiz; Carvalho, Márcio Luiz Bunte de
Resumo: The simulation of artificial neural networks (ANN) is a hard task due to two reasons: the complexity of ANN sprecification and the computational cost associate. The first problem has been addressed by some simulators by allowing specification of ANNs. Parallel processing has been recognized as a solition for the second problem. This work presents a neural network simulation strategy that combines different types of parallel processors and servers. This strategy is implemented in the HENNS, an environment that allows the user to describe and simulate various neural network models in parallel and heterogeneous environments. The results show that the use of heterogeneity is better than the use of parallel architectures alone.
Código DOI: 10.21528/CBRN1994-020
Artigo em PDF: CBRN1994-paper20.pdf
Arquivo BibTex: CBRN1994-paper20.bib