Electrical and Computer Engineering
Digital Signal Processing
Christopher Rozell (Rice MS '02, PhD ' 07)
Assistant Professor; Demetrius T. Paris Junior Professor Bioengineering, and Digital Signal Processing
Georgia Institute of Technology
Exploring optimal sensory coding theories for neural systems under biophysical constraints
Monday, March 24, 2014
to 5:00 PM
1049 George R. Brown Hall
6100 Main St
Houston, Texas, USA
Optimal sensory coding theories (e.g., sparse coding, predictive coding, divisive normalization, decision theory) have often been developed in an attempt to account for observed electrophysiology results in terms of high-level coding goal. While these types of models have achieved some success, it is less common to explore what insight is gained from mapping these models to biophysically realistic mechanisms. In this talk I will describe our recent work in two areas where optimal sensory coding models intersect with biophysical mechanisms. Most of the time will be spent in part 1, where we examine the sparse coding hypothesis as a model of primary visual cortex (V1). By using engineering approaches (e.g., low-rank matrix factorization, dynamical systems, optimization) to develop a biophysically realistic mechanistic model, I will describe our recent results showing how a wide range of experimental observations are emergent properties of this implementation. These results include nonlinear and nonclassical receptive field response properties (single cell and population), inhibitory interneuron receptive field properties, and active decorrelation of natural scene statistics. In part 2 we will examine the differences in activity propagation through sensory pathways from sensory and artificial stimulation. I will describe our recent efforts in the rat vibrissa pathway to combine these biophysical constraints with engineering approaches of optimal signal set design to develop effective strategies for surrogate sensory inputs from electrical or optogenetic artificial stimulation.
Host: Richard Baraniuk
Biography of Christopher Rozell (Rice MS '02, PhD ' 07):
Christopher J. Rozell received a B.S.E. degree in Computer Engineering and a B.F.A. degree in Music (Performing Arts Technology) in 2000 from the University of Michigan. He attended graduate school at Rice University, receiving the M.S. and Ph.D. degrees in Electrical Engineering in 2002 and 2007, respectively. Following graduate school he joined the Redwood Center for Theoretical Neuroscience at the University of California, Berkeley as a postdoctoral scholar. In 2008 Dr. Rozell joined the faculty at the Georgia Institute of Technology where he is currently an Assistant Professor and holds the Demetrius T. Paris Junior Professorship in Electrical and Computer Engineering.
His research interests live at the intersection of signal processing, machine learning and computational neuroscience. Specifically, his lab uses tools from modern data analysis to improve our understanding of neural systems and insight from modern neuroscience to build more effective computational systems, with applications ranging from biotechnology to remote sensing. His research lab is affiliated with both the Center for Signal and Information Processing and the Laboratory for Neuroengineering. Dr. Rozell received the National Science Foundation CAREER Award in 2014, and previously was the recipient of the Texas Instruments Distinguished Graduate Fellowship at Rice University. In addition to his research activity, Dr. Rozell was awarded the CETL/BP Junior Faculty Teaching Excellence Award at Georgia Tech in 2013.