Abstract In this paper, we present a new technique for mapping the backpropagation algorithm on hypercubes and related architectures. A key component of this technique is a network partitioning scheme which is called checkerboarding. Checkerboarding allows us to replace the all-to-all broadcast operation performed by the commonly used vertical network partitioning scheme, with operations that are much faster on the hypercubes and related architectures. Checkerboarding can be combined with the pattern partitioning technique to form a hybrid scheme which performs better than either one of these schemes. Theoretical analysis and experimental results on nCUBE2TM y and CM5TM z show that our scheme performs better than the other schemes, both for uniform and non-uniform networks.

Both from a mathematical as well as a biological perspective, Wavelet transforms present themselves as an attractive means for extracting low-level information from an image. We present a processor-time optimal algorithm to implement wavelet transforms on a VLSI implementable parallel digital architecture. 1 Why 2-D Wavelet Transforms? A desirable initial processing step in an artificial vision system is the identification of certain attributes in an incoming image. Traditionally, this is done by transforming the image into a domain where these attributes or features emerge as coefficients corresponding to a set of expansion functions. Transforms in computer vision have been used for image analysis, segmentation, feature extraction and representation, data compression and object recognition. Image transforms can be classified on the basis of the resolution they provide in the visual space domain versus the spatial frequency domain. While the pixel by pixel grey level descripition of t...

As artificial neural networks (ANNs) gain popularity in a variety of application domains, it is critical that these models run fast and generate results in real time. Although a number of implementations of neural networks are available on sequential machines, most of these implementations require an inordinate amount of time to train or run ANNs, especially when the ANN models are large. One approach for speeding up the implementation of ANNs is to implement them on parallel machines. This paper surveys the area of parallel environments for the implementations of ANNs, and prescribes desired characteristics to look for in such implementations. 1 Introduction Although traditional von Neumann computing has been successful in many applications, it has not proved effective in solving a variety of important complex problems. At the same time, it has been observed that human beings solve these problems routinely in real time. Typical problems that fall into this class consist of perception...

An important step in image processing tasks involves the identification of certain desired attributes in an image. Traditionally, this is done by transforming the image into a domain where the desired attributes or features are easily identifiable. In this paper, we discuss the parallel implementation of one such image transform--the 2-D Gabor based Wavelet Transform. Individual components of this transform are sensitive to particular ranges of frequencies and the orientation of features in an image. The transform is formed by computing convolutions of the image with a family of wavelets. Each member of the wavelet family is a 2-D Gabor Function. We describe how this 2-D Wavelet Transform can be computed efficiently on a fine-grained, Single Instruction, Multiple Data Stream (SIMD) computer, the Connection Machine (CM-2). The transform of a 128 \Theta 128 pixel image using 40 wavelets (sensitive to different frequency levels and orientations of features) takes 2.43 seconds on the CM-2 ...

Abstract: Multiple-Layer Perceptrons (MLPs) running Backpropagation are still a very frequently utilized artificial neural networks paradigm. Particularly when applied to very highdimensional data sets leading to rather large networks, their training may take too long. Besides some architectural or solely numerical modifications, especially a parallel implementation is suitable to speed up the network training. This paper evaluates a Message Passing Interface (MPI) based parallel variant of Backpropagation which was run on a number of different parallel computer architectures. A standard character recognition problem with a wavelet transform based feature data set of 262 input dimensions is applied as benchmark environment.

An important step in image processing tasks is the identification of certain desired attributes in an image. Traditionally, this is done by transforming the image into a domain where the desired attributes or features are easily identifiable. One class of image transforms that have found great applicability in artificial vision systems are those that transform the image from the spatial to the frequency domain (the most popular of these is the Fourier Transform). It is desirable to use a member of this class that yields good resolution in both the spatial as well as the frequency domains. In this paper, we consider a transform called the 2-D Gabor Wavelet Transform that achieves the best possible conjoint resolution in space and frequency. This transform is based on families of functions that are translated, rotated or dilated versions of each other. In addition to their desirable mathematical properties, research in psychophysics has shown that the responses of simple cells in mammali...

Abstract not found