For many real-world pattern learning and classification tasks, today’s digital computers cannot compete with human brains in terms of energy and computational efficiency. One of the reasons is that the computer architecture is very different from that of our neural systems, in which a huge number of nerve cells communicate with action potentials, which we call ‘spikes’, in parallel. Human brain works with spikes and biological spiking neurons. Spiking neural network is biologically inspired and grounded under a solid scientific framework to achieve some of the advantages of biological systems. It has been a general belief that spiking neural network is the computational way to achieve brain like performance. The research on the theory and implementation of Spiking Neural Network forms the foundation of spiking neural network computing.
Leveraging on the research outcomes in the past decades, this project is focused on developing realistic spiking neural network framework that is hardware implementable by modeling the following brain parts: synapses (plasticity, learning), visual cortex (encoding and recognition), hippocampus (memory and synthesis). The framework will include neural encoding, precise-spike-driven learning for synaptic plasticity learning, neuron modeling for analog memory element characteristics, and classification and synthesis of spatiotemporal patterns.
Project Duration: 31 March 2017 – 30 March 2022
Funding Source: RIE2020 Advanced Manufacturing and Engineering Programmatic Grant A1687b0033
Acknowledgement: This research work is supported by Programmatic Grant No. A1687b0033 from the Singapore Government’s Research, Innovation and Enterprise 2020 plan (Advanced Manufacturing and Engineering domain). Project Title: Neuromorphic Computing.