S. Churcher

Integrated pulse stream neural networks: results, issues, and pointers

Results from working analog VLSI implementations of two different pulse stream neural network forms are reported. The circuits are rendered relatively invariant to processing variations, and the problem of cascadability of synapses to form large systems is addressed. A strategy for interchip communication of large numbers of neural states has been implemented in silicon and results are presented. The circuits demonstrated confront many of the issues that blight massively parallel analog systems, and offer solutions.

Pulse stream VLSI neural networks

EPSILON, a large, working, VLSI device, demonstrates pulse stream methods in the wider context of analog neural networks. EPSILON uses dynamic weight storage techniques, but a nonvolatile alternative is desirable. To that end, we have developed an amorphous silicon memory, which we present in experiments incorporating the device in a modest pulse stream neural chip. We have also developed a target-based training algorithm, which we demonstrate in a prototype learning device using a realistic problem. Finally, we explore system-level problems in experiments with a second version of EPSILON in a small, autonomous robot.<<ETX>>

Pulse-Firing VLSI Neural Circuits for Fast Image Pattern Recognition

Neural networks are identified as an effective means of performing pattern recognition on image data. To take advantage of the parallelism inherent in such architectures, an analogue CMOS VLSI circuit has been developed. This device consists of novel pulse-firing neurons and synapses. The neurons are current controlled oscillators, which deplete their own input activity whenever a pulse is fired, in a manner similar to biological neurons. The synapses use neural voltage pulses to create pulsed current outputs. The magnitude of these outputs is determined by the synaptic weight voltage. When configured as a multi-layered perceptron, the neural circuit will facilitate the “real-tune” labelling of regions in segmented images.