Title :
Back propagation in realistic parallel environment
Author :
Strupl, David ; Neruda, Roman
Author_Institution :
Fac. of Math. & Phys., Charles Univ., Prague, Czech Republic
Abstract :
Various ways of parallelizing the back propagation algorithm (BP) are studied on the bulk-synchronous parallel machine (BSP) like parallel model. Optimal number of processors and criteria for choosing the right task decomposition are presented. Theoretical results are verified in the parallel virtual machine (PVM) environment on a cluster of workstations
Keywords :
backpropagation; neural net architecture; optimisation; virtual machines; BSP-like parallel model; PVM environment; back propagation; bulk-synchronous parallel machine; neural nets; parallel environment; parallel virtual machine; task decomposition; workstation cluster; Clustering algorithms; Computational modeling; Computer networks; Concurrent computing; Hardware; Multilayer perceptrons; Neural networks; Neurons; Parallel machines; Workstations;
Conference_Titel :
Intelligence and Systems, 1998. Proceedings., IEEE International Joint Symposia on
Conference_Location :
Rockville, MD
Print_ISBN :
0-8186-8548-4
DOI :
10.1109/IJSIS.1998.685414
