Professor Constantine Caramanis was awarded an NSF CAREER award, for his work at the intersection of robust optimization, networks and machine learning. The award will support his work on 'High Dimensional Statistics - Adaptive Networks, Structure and Robustness,' with the goal of developing a new methodological framework to understand high dimensional complex phenomena from potentially corrupted and incomplete data. This research has the potential to significantly impact our abilities to discover important structure in complex problems, from bioinformatics to social networks. As high-dimensional data become increasingly pervasive (the length of genomes sequenced increases; the number of patients carrying a genetic disease does not) computationally scalable techniques to analyze and mine data, finding structure and patterns, may hold the key to important scientific progress. Developing novel approaches using robust optimization, this work will develop an efficiently scalable algorithmic framework to attack such problems, and address issues of corrupted or missing data. One major application thrust will be in user behavior prediction to help better understand how to provision large-scale engineered systems. One important example is transmitting video over wireless networks, where being able to predict user demand can be instrumental in helping to smooth out the peaks and valleys, allowing the user experience to improve far beyond how much capacity is expanded.
