When
Spiking Algorithms for Neuromorphic Hardware
Neural networks have exploded in popularity in the last decade. While they have achieved widespread success across many disciplines, they typically require massive amounts of data, time, and energy to train. Originally inspired by biological neural networks in the brain, it turns out these artificial neural networks are not very biologically realistic, and their implementation on conventional computing platforms fails to realize the parallel computing capabilities of neurons. Neuromorphic computing is an emerging field that takes inspiration from the brain in the form of hardware and algorithms that emulate the massively parallel and event-based processing capabilities of more biologically realistic spiking neural networks. In this talk I will discuss neuromorphic hardware and spiking neural networks, and then describe some of the work myself and others at Sandia are doing with spiking algorithms on neuromorphic hardware.
Place: Hybrid: Math, 402/Zoom: https://arizona.zoom.us/j/83541348598 Password: BB2022