Electrical and Computer Engineering
Digital Signal Processing
Group Leader, Janelia Farm Research Campus
Howard Hughes Medical Institute
ECE Neuroengineering Seminar Series
Thursday, September 5, 2013
What does a neuron do? An online signal processing perspective
to 5:00 PM
1064 George R. Brown Hall
6100 Main St
Houston, Texas, USA
A biological neuron is an elementary building block of the brain. While the physiology of a neuron has been studied extensively its computational role has until now remained a mystery. We propose to view a neuron as a signal processing device for extracting and tracking a non-Gaussian feature from its high-dimensional input. Such task can be accomplished collectively by two online algorithms: a slow time-scale algorithm which adjusts synaptic weights to extract the most non-Gaussian projection of the high-dimensional input, and a fast time-scale algorithm which tracks, or de-noises, the projection amplitude. Both online algorithms rely on sparsity-inducing regularizers and have provable performance bounds. The steps of these algorithms account for the salient physiological features of neurons such as leaky integration, non-linear output function, Hebbian synaptic plasticity rules, sparse connectivity and activity. Therefore, our work sets up a foundation for modeling biological neurons as adaptive signal processing devices.
Host: Ashok Veeraraghavan
A reception will follow in Duncan Hall 1070.
Biography of Dmitri Chklovskii:
Dmitri Chklovskii studied physics and engineering in St. Petersburg, Russia, then obtained a PhD in theoretical physics from MIT in 1994. After being a Junior Fellow at the Harvard Society of Fellows he switched to theoretical neurobiology and was a Sloan Fellow at the Salk Institute. In 1999, he founded a theoretical neuroscience group at Cold Spring Harbor Laboratory, where he was an Assistant and then Associate Professor. In 2007 he moved to Janelia Farm Research Campus of the Howard Hughes Medical Institute as a Group Leader. Chklovskii’s work is highly interdisciplinary, applying ideas from electrical engineering, computer science, applied math and physics to reverse engineer the brain.