Skip to main content.
__ __

Magnet Photo

Design and Application of Neuromorphic Chips

Abstract: I'll talk about a current focus of the neural engineering research in my laboratory: Making it easy to use neuromorphic chips in a large variety of applications. The Neural Engineering Framework (NEF) is a formal method for mapping arbitrary computations onto networks of spiking neurons by programming their synaptic strengths. For example, you can get spiking neurons to perform temproal integration or to rotate a two-dimensional vector by a given angle. Why would you want to use NEF instead of the standard paradigm for training neural networks? Well, in cases where you have a mathematical solution (e.g., motor control), this is a much easier way to get a neural network to do what you want than the standard paradigm, which requires tons of training examples (inputs and corresponding outputs). And why would you want to do this with spiking neurons rather than the standard rate neurons? Because you could ran your algorithm in real time and at low power on neuromorphic hardware. I'll present our results mapping a Kalman-filter-based brain-machine interface onto a spiking neural network and controlling a three-degree-of-freedom robot arm with a neuromorphic chip. I'll close with an overview of next-generation neuromorphic chip designs that will integrate a million spiking neurons with a billion programmable synaptic connections on a single chip that dissipates milliwatts of power.
Speaker: Kwabena Boahen - Stanford University
Speaker Bio: Kwabena Boahen is Professor of Bioengineering at Stanford University, where he directs the Brains In Silicon Laboratory. He is a bioengineer who is using silicon integrated circuits to emulate the way neurons compute, linking the seemingly disparate fields of electronics and computer science with neurobiology and medicine. His contributions to the field of neuromorphic engineering include a silicon retina that could be used to give the blind sight and the first system that simulates a million neurons—with billions of synaptic connections—in real-time. His scholarship is widely recognized, with over eighty publications to his name, including a cover story in Scientific American (May 2005). His TED talk, “A computer that works like the brain” (2007), has been viewed half-a-million times. He has received several distinguished honors, including the National Institutes of Health Director’s Pioneer (2006) and Transformative Research (2011) Awards. Esquire Magazine recognized him—along with fifteen other distinguished researchers—in The Brightest: 16 Geniuses That Give us Hope (2010). Professor Boahen’s BS and MSE degrees are in Electrical and Computer Engineering (both earned in 1989 from Johns Hopkins University, Baltimore MD). His PhD degree is in Computation and Neural Systems (earned in 1997 from the California Institute of Technology, Pasadena CA). Before moved to Stanford, he was on the faculty of the University of Pennsylvania, Philadelphia PA (1997 to 2005).
Poster Link: Poster