TOPIC:                                   Towards a Global Brain Model:  Understanding Cortical Maps and Temporal Dynamics

GUEST SPEAKER:               Dr. Ravi Rao, IBM TJ Watson Research Center

DATE:                                     Monday, November 20, 2006

TIME:                                      2:00PM

LOCATION:                           Harris Center, Room 101

As part of our program of informing the local community of the School of Electrical Engineering & Computer Science Colloquia, I am forwarding you the announcement for the next Colloquium.  Please pass this announcementon to members of your organization who might be interested in attending.

For those of you who are not familiar with the campus, please go to the Visitor Information Booth across from the Administration Building and receive parking instructions.

Information about the School of Electrical Engineering & Computer Science can also be found on Home Page



In the field of computational neuroscience, data are gathered at multiple scales of abstraction, from the individual neuronal level to whole brain signals such as from functional magnetic resonance imaging. In order to better interpret these data, it is necessary to develop a global perspective on how the vertebrate brain operates. This is a challenging problem, and I will present the research done over the past five years at IBM Research in addressing it.
Specifically, we will examine an important property of the visual cortex, which exhibits self-organized two dimensional maps representing several visual cues such as orientation, ocular dominance, frequency and motion. A computational model that explains the formation of these maps will be presented.
Another significant aspect of neural computation is the presence of temporal dynamics, which can be modeled by oscillatory networks.  We will present a novel network of oscillatory units, whose behavior is described by the amplitude and phase of oscillations. We derive the network dynamics from an objective function that rewards both the faithfulness and the sparseness of representation. The resulting network architecture is simple, and the dynamics are straightforward to interpret.   Network units exhibit synchronization through phase locking after an initial settling period. The nature of phase synchronization is such that it binds specific input units with the output units that represent a classification of these input units. The significance of this binding to global brain modeling is that it addresses a problem the brain must solve: the integration of multiple visual cues represented in different areas of the brain.

Short Bio:

Dr. Ravishankar Rao is a Research staff member with the Biometaphorical Computing group at the IBM T.J. Watson Research Center, Yorktown Heights, New York. His research interests include image processing, computer vision and computational neuroscience. He is an Associate Editor of the journals Pattern Recognition and Machine Vision and Applications. He received his B.Tech degree in electrical engineering from the Indian Institute of Technology (Kanpur) in 1984, and Ph.D. degrees in computer engineering from the University of Michigan, Ann Arbor, in 1989 respectively. His work has resulted in fifteen patents and over forty publications. He has published a book entitled ``A Taxonomy for Texture Description and Identification.'' He is a Senior Member of the IEEE and a member of the Society for Neuroscience. He was named a Master Inventor at IBM Research in 2004, in recognition of his sustained invention activity resulting in valuable patents within IBM’s portfolio.