J. García, E. Martínez, A. Arquero, and C. Gonzalo (Spain)
Neural networks, parallel processing,
message-passing model, pattern characteristic, remote
sensing image classification.
Four different parallel approaches for computing the
operation phase of Fritzke Neural Networks (NNs) have
been proposed, investigated and compared in this paper.
All of them are based on a multicomputer system, by
applying the message-passing model. The real problem of
remote sensing image classification has been considered
for evaluation sake. From results, it can be concluded that
it is possible to obtain promising speedups and efficiency
with a suitable number of processors.