Abstract
We present an implementation of Kohonen Self-Organizing Feature Maps for the Spert-II vector microprocessor system. The implementation supports arbitrary neural map topologies and arbitrary neighborhood functions. For small networks, as used in real-world tasks, a single Spert-II board is measured to run Kohonen net classification at up to 208 million connections per second (MCPS). On a speech coding benchmark task, Spert-II performs on-line Kohonen net training at over 100 million connection updates per second (MCUPS). This represents almost a factor of 10 improvement compared to previously reported implementations. The asymptotic peak speed of the system is 213 MCPS and 213 MCUPS.
Preview
Unable to display preview. Download preview PDF.
References
Wawrzynek, J., Asanović, K., Kingsbury, B. E. D., Beck, J., Johnson, D., Morgan, N.: Spert-II: A vector microprocessor system, IEEE Computer, 29(3):79–86, March 1996.
Kohonen, T.: Self-organizing formation of topologically correct feature maps, Biological Cybernetics, 43(1):59–69, 1982.
Myklebust, G., Solheim, J. G.: Parallel self-organizing maps for actual applications, Proceedings of the IEEE International Conference on Neural Networks, Perth, 1995.
Cornu, T., Ienne, P.: Performance of digital neuro-computers, Proceedings Fourth International Conference on Microelectronics for Neural Networks and Fuzzy Systems, September 1994, 87–93.
Asanović, K., Johnson, D.: Torrent architecture manual, Technical Report, Computer Science Division, University of California at Berkeley, CSD-97-930, 1997.
Kane, G.: MIPS RISC Architecture (R2000/R3000), Prentice Hall, 1989.
Russel, R. M.: The CRAY-1 computer system, Communications of the ACM, 21(1):63–72, January 1978.
Asanović, K., Beck, J.: T0 engineering data, Technical Report, Computer Science Division, University of California at Berkeley, CSD-97-931, 1997.
Ienne, P., Cornu, T., Kuhn, G.: Special-purpose digital hardware for neural networks: An architectural survey, Journal of VLSI Signal Processing, 13(1):5–25, 1996.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1997 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Asanović, K. (1997). A fast Kohonen net implementation for spert-II. In: Mira, J., Moreno-DÃaz, R., Cabestany, J. (eds) Biological and Artificial Computation: From Neuroscience to Technology. IWANN 1997. Lecture Notes in Computer Science, vol 1240. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0032538
Download citation
DOI: https://doi.org/10.1007/BFb0032538
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-63047-0
Online ISBN: 978-3-540-69074-0
eBook Packages: Springer Book Archive