On-chip learning in pulsed silicon neural networks

Self-learning chips to implement conventional ANN (artificial neural network) algorithms are very difficult to design and unconvincing in their results. We explain why this is so and say what lessons previous work teaches us in the design of self-learning systems. We offer an alternative, 'biologically-inspired' approach, explaining what we mean by this term and providing an example of a robust, self-learning design which can solve simple classical-conditioning tasks.

[1]  Chih-Ming Ho,et al.  Analog VLSI system for active drag reduction , 1996, Proceedings of Fifth International Conference on Microelectronics for Neural Networks.

[2]  J. J. Paulos,et al.  On-chip learning in the analog domain with limited precision circuits , 1992, [Proceedings 1992] IJCNN International Joint Conference on Neural Networks.

[3]  Lloyd W. Massengill,et al.  Weight decay and resolution effects in feedforward artificial neural networks , 1991, IEEE Trans. Neural Networks.

[4]  Tetsuro Itakura,et al.  Neuro chips with on-chip back-propagation and/or Hebbian learning , 1992 .

[5]  Takashi Morie,et al.  An all-analog expandable neural network LSI with on-chip backpropagation learning , 1994, IEEE J. Solid State Circuits.

[6]  Heekuck Oh,et al.  Neural Networks for Pattern Recognition , 1993, Adv. Comput..

[7]  A. Klopf.  A neuronal model of classical conditioning , 1988 .

[8]  Torsten Lehmann,et al.  Nonlinear backpropagation: doing backpropagation without derivatives of the activation function , 1997, IEEE Trans. Neural Networks.

[9]  P. Fromherz,et al.  Silicon-Neuron Junction: Capacitive Stimulation of an Individual Neuron on a Silicon Chip. , 1995, Physical review letters.

[10]  Torsten Lehmann,et al.  Mixed analog/digital matrix-vector multiplier for neural network synapses , 1996 .

[11]  David P. M. Northmore,et al.  Switched-capacitor neuromorphs with wide-range variable dynamics , 1995, IEEE Trans. Neural Networks.

[12]  Torsten Lehmann.  ECCOPUNCH: the Edinburgh classical conditioning pulsed neural chip , 1997, Proceedings of 1997 IEEE International Symposium on Circuits and Systems. Circuits and Systems in the Information Age ISCAS '97.

[13]  Davide Badoni,et al.  LANN27: an electronic implementation of an analog attractor neural network with stochastic learning , 1995, SPIE Defense + Commercial Sensing.

[14]  T. Lehmann.  Teaching pulsed integrated neural systems: a psychobiological approach , 1996, Proceedings of Fifth International Conference on Microelectronics for Neural Networks.

[15]  Marwan A. Jabri,et al.  Weight perturbation: an optimal architecture and learning technique for analog VLSI feedforward and recurrent multilayer networks , 1992, IEEE Trans. Neural Networks.

[16]  John Wawrzynek,et al.  Systems technologies for silicon auditory models , 1994, IEEE Micro.

[17]  Ronald S. Gyurcsik,et al.  Building blocks for a temperature-compensated analog VLSI neural network with on-chip learning , 1994, Proceedings of IEEE International Symposium on Circuits and Systems - ISCAS '94.

[18]  Davide Badoni,et al.  Electronic implementation of an analogue attractor neural network with stochastic learning , 1995 .

[19]  Patrick A. Shoemaker,et al.  A hierarchical clustering network based on a model of olfactory processing , 1992 .