Spin-L: sequential pipelined neuroemulator with learning capabilities

In this paper, an extensive study of learning and retrieval algorithms for Hopfield's pattern classifier network, multilayer backpropagation, Kohonen's self-organized feature mapping network, and the binary adaptive resonance theory (ART-1) models is reported. Parallelism as well as computational requirements are identified for all algorithms. The algorithms are then mapped onto the sequential pipelined neuroemulator (SPIN) architecture. As a result, the SPIN with learning (SPIN-L) machine is developed as an enhanced architecture to accommodate the new requirements.

[1]  Stamatis Vassiliadis,et al.  SPIN: a sequential pipelined neurocomputer , 1991, [Proceedings] Third International Conference on Tools for Artificial Intelligence - TAI 91.

[2]  J J Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities. , 1982, Proceedings of the National Academy of Sciences of the United States of America.

[3]  Bernard Widrow,et al.  Adaptive switching circuits , 1988 .

[4]  T. Kohonen.  Self-Organized Formation of Correct Feature Maps , 1982 .

[5]  D. O. Hebb,et al.  The organization of behavior , 1988 .

[6]  Jenq-Neng Hwang,et al.  A Unified Systolic Architecture for Artificial Neural Networks , 1989, J. Parallel Distributed Comput..

[7]  Stephen Grossberg,et al.  A massively parallel architecture for a self-organizing neural pattern recognition machine , 1988, Comput. Vis. Graph. Image Process..