Artificial neural networks on reconfigurable meshes
Artificial neural networks(ANN) have been used successfully in applications such as pattern recognition, image processing, automation and control. Majority of today's applications use backpropagate feedforward ANN. In this paper, two methods of P pattern L layer ANN learning on n x n RMESH have been presented. One required memory space of O(nL) but conceptually is simpler to develop and the other uses pipelined approach which reduces the memory requirement to O(L). Both of these algorithms take O(PL) time and are optimal for RMESH architecture.
KeywordsArtificial Neural Networks Reconfigurable Mesh Algorithms Parallel Algorithms
Unable to display preview. Download preview PDF.
- J. Chung, H. Yoon, and S. R. Maeng, “A systolic Array Exploiting the Inherent Parallelisms of Artificial Neural Networks”, International Conference on Parallel Processing, vol. 1, pp 652–653, 1991Google Scholar
- S. Haykin, Neural Networks, A Comprehensive Foundation, IEEE Press, 1994Google Scholar
- A. Hiraiwa, S. Kurosu, S. Arisawa and M. Ionue, “A Two Level pipeline RISC Processor Array for ANN”, International Joint Conference on Neural Networks, Washington DC, vol. 2, ppl37–140, 1990Google Scholar
- .J Jenq and S. Sahni “Reconfigurable Mesh Algorithms for Fundamental Data Manipulation Operations”, Computing on Distributed Memory Multiprocessors, NATO Series F, ed. F. Ozguner, Springer Verlag, 1993Google Scholar
- J. Jenq and W. Li “Artificial Neural Networks on Reconfigurable Meshes”, CSCI-TR-98-01 Department of Computer Science, University of ArkansasGoogle Scholar
- S. Y. Kung, “Parallel Architectures for Artificial Neural Nets”, International Conference on Systolic Arrays, pp 163–174, 1988Google Scholar
- K. Parker and A. Thornbrugh, “Parallelized Back-Propagation Training and Its Effectiveness”, vol. 2, International Conference on Neural Networks, Washington DC, vol. 2, pp, 179–182, Jan. 1990Google Scholar