Genetic algorithms in parallel neural networks

In this paper we show a Parallel Neural Network (Cross-over Prediction Model) for time series forecasting implemented in PVM (#x0201C;Parallel Virtual Machine”) and MPI (“Message Passing Interface”), in order to reduce computational time. Pamlleliration is achieved twofold: (a) updating autoregressive paramctcrs rising a genetic algorithm (GA) and (hi evaluating the overall prediction function via a parallel neural network. We implement the GA in two popular architectures of parallel processors (i.e. hypercube and 2D-mesh) and discuss their time efficiency.