Short- and long-term dynamics in a stochastic pulse stream neuron implemented in FPGA
This paper presents the implementation of the Hebbian learning rule in a hardware-friendly architecture based on a stochastic pulse representation of the signals. We compare implementation costs and speed of this approach with those of a parallel and a bit-serial implementation. The neural model includes both short- and long-term dynamics. Hence, networks composed of these neurons solve delayed reinforcement and adaptive timing tasks which has been shown in several real-world applications.
Unable to display preview. Download preview PDF.
- 1.A. Bühlmeier, G. Manteuffel, M. Rossmann, and K. Goser. Robot learning in analog neural hardware. In Artificial Neural Networks: 6th international conference; proceedings (ICANN 96), Bochum, Germany, volume 1112 of Lecture Notes in Computer Science, pages 311–316. Springer-Verlag Berlin, July 16–19 1996.Google Scholar
- 2.A. Bühlmeier, G. Manteuffel, M. Rossmann, and K. Goser. Application of a local learning rule in a wheelchair robot. In Third International Conference on Neural Networks and their Applications (Neurap97), pages 177–182, March 12–13 1997.Google Scholar
- 3.J. Hertz, A. Krogh, and R. Palmer. Introduction to the Theory of Neural Computation. Addison-Wesley, 1991.Google Scholar
- 4.P. D. Hortensius, M. R. D., and H. C. Card. Parallel random number generation for vlsi systems using cellular automata. IEEE Transactions on Computers, 38:1466–1473, April 1989.Google Scholar
- 5.L. Massen. Stochastische Rechentechnik. Carl Hanser Verlag, 1977. in German.Google Scholar
- 6.A. Murray and L. Tarassenko. Analogue Neural VLSI — A Pulse Stream Approach. Chapman & Hall, London, UK, 1994.Google Scholar
- 7.M. Rossmann and K. Goser. Dynamic artificial neural networks for adaptive control of velocity of a track vehicle. Technical report, Department of Microelectronics, Faculty of Electrical Engineering, University of Dortmund, Germany, 1997. to be published.Google Scholar
- 8.M. Rossmann, T. Jost, K. Goser, A. Bühlmeier, and G. Manteuffel. Exponential hebbian on-line learing implemented in FPGAs. In Artificial Neural Networks: 6th international conference, proceedings (ICANN 96), Bochum, Germany, volume 1112 of Lecture Notes in Computer Science, pages 767–772. Springer-Verlag Berlin, July 16–19 1996.Google Scholar
- 9.E. Vittoz. Analog VLSI signal processing: Why, where and how. Journal of VLSI Signal Processing, 8:27–44, July 1994.Google Scholar
- 10.N. Wiener. Cybernetics. MIT Press, Cambridge, Massachusetts, 1948.Google Scholar