News

Embedded systems often operate on sensitive data in safety-critical environments, including transportation, health care, and industrial control. Embedded software can leak information about the activity and data via physical side-channels, such as electromagnetic (EM) fields, which can be measured with ease by an adversary using modest equipment. Such measurements can be used to profile programs, find anomalies in the software, identify sensitive information, and most fundamentally, reveal what instructions are being executed on the system.

Prof. Joydeep Ghosh of UT ECE was the keynote speaker at the inaugural Workshop on Divergences and Divergence Learning (WDDl), held in Atlanta, June 2013. In his talk, entitled "Learning Bregman Divergences for Prediction with Generalized Linear Models," which reflects joint work with ECE and WNCG student Sreangsu Acharrya,  an efficient approach to learning a broad class of predictive models was introduced. What is most remarkable about this approach is that model parameters can be estimated even when the loss function is unknown.