As the next generation of neural networks, Spiking Neural Networks feature powerful and energy-efficient computation. In combination with the hardware specifically optimized for SNN workloads, further improvement of computation and energy efficiency of SNN can be achieved.
This seminar focuses on the design of specific hardware for spike-based information processing, and will cover the following topics:
1. Neurophysiological Background for Spiking Neural Network: human brain, synaptic transmission, excitatory and inhibitory synapse, action potential, synaptic integration, refractory period, long-term potentiation and long-term depression, spike timing dependent plasticity, lateral inhibition.
2. Spiking Neural Network: Hodgkin-Huxley model, Izhikevich model, leaky-integrate-and-fire model, spike-timing-dependent plasticity, winner-take-all, spiking neural network
3. Neuromorphic Processor Design: digital logic design basics, ASIC design flow, neuromorphic processors.
Some paper review presentation tasks are included as part of the seminar. A technical report summarizing the presented papers is required at the end of the semester.
- Kursleiter:in: Dedong Zhao