Spiking neural network for feature recognition: Modelling and optimization.
Ph.D. Student Chunming Jiang
Department of Mechanical Engineering, University of Canterbury
Time & Place
Wed, 19 May 2021 11:00:51 NZST in Jack Erskine 443 (Lecture Theatre)
Spiking Neural Networks (SNNs) have recently attracted significant research interest as the third generation of artificial neural networks that can enable low-power event driven data analytics. Spiking neurons modulate the nervous cells via receiving external incentives, generation of action potential and firing spikes, which is more biologically plausible. With less energy and faster execution, SNNs have broad prospects for feature recognition on the specific neuromorphic hardware, such as signals monitoring of smart watches, robots grab objects.
We adopted a two-layer SNN with unsupervised learning to recognize tactile signals. This method is simpler than existing methods and show a good accuracy. Also, to decrease the accuracy gap of ANNs and SNNs, we proposed a new spiking neural model KLIF. The result shows a promoted and comparable accuracy compared with ANN while using KLIF with the supervised learning to train a multi-layer SNN.
Supervisor: Associate Prof Yilei Zhang
Chunming Jiang is a PhD student at the University of Canterbury (UC). His research interests include the optimization of bio-inspired spiking neural networks, object recognition.