Related Experiment Videos
SPIKEII: an integrate-and-fire AVLSI chip
L S Smith1, B E Eriksson, A Hamilton
1Department of Computing Science and Mathematics, University of Stirling, Scotland.
International Journal of Neural Systems
|January 12, 2000
Summary
The SPIKEII chip is a new analogue VLSI neural network chip. It features programmable synapses and is designed for integrate-and-fire neural networks.
Area of Science:
- Neuroscience
- Computer Engineering
- Artificial Intelligence
Background:
- The development of efficient hardware for artificial neural networks is crucial for advancing AI.
- Analogue Very-Large-Scale Integration (VLSI) offers potential for low-power, high-speed neuromorphic computing.
Purpose of the Study:
- To introduce the SPIKEII chip, an advancement in analogue VLSI neural network hardware.
- To detail the circuitry and functionality of programmable synapses in the SPIKEII chip.
- To demonstrate the chip's performance with integrate-and-fire neural networks.
Main Methods:
- Design and fabrication of the SPIKEII chip using analogue VLSI technology.
- Implementation of programmable synapses for flexible neural network configurations.
- Testing of neural networks composed of integrate-and-fire neurons on the SPIKEII chip.
Main Results:
- Successful implementation of the SPIKEII chip, building upon the SPIKEI architecture.
- Demonstration of programmable synapse functionality within the analogue VLSI framework.
- Validation of the chip's capability to simulate networks of integrate-and-fire neurons.
Conclusions:
- The SPIKEII chip represents a significant step in creating efficient, hardware-based neural networks.
- Analogue VLSI with programmable synapses is a viable approach for neuromorphic computing.
- Further research can explore advanced applications using the SPIKEII platform.