Physical Sciences and Mathematics Commons^™

Optimizing Communication Routing In Parallel Computer Architectures, Nikolaos T. Liolios

Masters Theses

This thesis is a research effort to improve on existing algorithms for processor placement and routing on hypercube and 2d-transputer parallel architectures that perform communications through message propagation. Given a set of processors N, randomly distributed on the parallel computer network, the problem is to find an optimal communication routing that minimizes the number of required links and intermediate processors. It was concluded that the spanning tree is the best starting heuristic for the routing problem. For hypercubes an algorithm was developed that outperforms all algorithms in the literature. For 2d-transputers the method of overlap maximization used in VLSI routing, …

Signal Processing With Neural Networks, Pero Smrzlic

Masters Theses

In this study, we introduce the Adaptive Back Propagation (ABP) learning algorithm, computationally superior to the standard Back Propagation. The ABP is based on the new activation function, with its corresponding adaptive learning parameter. By using the combination of computer simulations and analysis in the domain of activation function, the Method of One Hidden layer was developed for the effective utilization of units in one-hidden layer networks. A parallel version of the ABP was designed and implemented on a nCUBE-2 supercomputer with 128 processors.

The simulation results suggested a strong correlation between frequency of signals and the role of hidden …