[Colloquium] Computer Science Seminar - Krishna V. Palem

[Donna Brooms]

COMPUTER SCIENCE DEPARTMENT SEMINAR
 
Monday, February 3, 2014
3:00 p.m.
Ryerson 251

Dr. Krishna V. Palemo
Rice University
www.cs.rice.edu/~kvp1
 
Title: “Sensoptimized Systems for ``Good enough’’ Computing
Ultra-efficient Cortical Processors through Melding Neuroscience with Inexact Architectures”.
 
Abstract: Increasingly, information systems such as cellphones, iPods and glasses—more broadly, embedded systems—are delivering information to be consumed by our senses. Such information, in the form of speech, graphics, or video, is subject to varying levels of processing by our nervous systems, followed by our higher cognitive functions in the brain.  Yet, system designs today do not often take advantage of the compensatory processing done neuro-cognitively by our brain. Rather, the current hardware, software, and industrial design methodologies aim to deliver the best possible quality to maximize the users’ positive experience. Often, the resulting high quality products are over-engineered and expensive—in terms of monetary cost, and the amount of energy (or battery) consumed. Over  several years now, we have been developing a philosophy and a design methodology to counter this trend—specifically aimed at the innovation of  digital computing systems which, when interacting with our senses, are optimized to be just “good enough” and thus not over-engineered This is achieved by factoring in the compensatory  neuro-cognitive processing done by our sensory pathways, and by trading away the accuracy of the system in return for disproportionately high savings or gains. The resulting sensoptimied  systems  are meant to be significantly more efficient than those designed conventionally. In this talk, will showcase a sensoptimized  good enough hearing-aid which we have been developing, to illustrate this principle.  At their core, our sensoptimized systems are realized using  inexact integrated circuits (ICs) and computing architectures, sometimes dubbed probabilistic CMOS (PCMOS)—a technology and design methodology which our group has been developing for over a decade—which we have shown can yield over an order of magnitude in energy savings, by trading accuracy and admitting tolerable error.  Looking into the future, inexact circuits and sensoptimization could be the basis for realizing families of cortical processors which meld principles of neuroscience with the design of good-enough computing platforms. Here, the opportunities are many and we will conclude the technical portion of our talk with an overview of a sensoptimized  corticial processor we are currently developing for realizing computer vision in real-time, based on architectures for spatio-temporal deep learning—whereas the neurons can be realized using inexact CMOS platforms, the synaptic delays are to be implemented using exciting and emerging magnetic spin-tourque memories. We will argue that sensoptimization has the potential for being a disruptive innovation in Clayton Christensen’s sense, with significant potential for impact at the bottom-of-the-pyramid and in emerging economies.
 
Krishna V. Palem is the Ken and Audrey Kennedy Professor at Rice University with appointments in Computer Science, in Electrical and Computer Engineering, and Statistics, and is a scholar in the Baker Institute for Public Policy. He founded and directed the NTU-Rice Institute on Sustainable and Applied Infodynamics (2010—2013). He was a Moore Distinguished Faculty Fellow at Caltech (2006—2007), and a Schonbrunn Fellow at the Hebrew University of Jerusalem (1999), where he was recognized for excellence in teaching. His advisee Suren Talla was awarded the Janet Fabri Prize for outstanding dissertation (2001), and his related work on the foundations of architecture assembly for designing reconfigurable embedded SoC architectures, developed at Proceler Inc. which he co-founded as a CTO, was a nominee (2002) for the Analysts choice awards as one of the four outstanding technologies.

A decade ago, he pioneered a novel technology dubbed Probabilistic CMOS (PCMOS) for enabling ultra low-energy computing through sensoptimization. In this area, his student Lakshmi Chakrapani's research was recognized through an outstanding dissertation award by Sigma Xi  (2008). PCMOS has also been recognized by a best-paper award at the IEEE-ACM CASES 2006 conference in Seoul, as one of the ten technologies 'likely to change the way we live' by MIT's Technology Review (2008), and as one of the seven 'emerging world changing technologies' by IEEE as part of its 125th anniversary celebrations (2009). More broadly, he has led efforts internationally in the area of embedded systems and their compiler optimizations, for which he has been named a Fellow of the IEEE, the ACM and AAAS. In 2012, Forbes (India) ranked him second on the list of eighteen scientists who are ``..some of the finest minds of Indian origin.” He is the recipient of the 2008 W. Wallace McDowell Award, IEEE Computer Society's highest technical award and one of computing's most prestigious individual honors.
 
Hosts: Prof. Andrew Chien
*Refreshments will be served prior to the talk at 2:00 in Ryerson 255*
 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20140129/6627feae/attachment.htm