Parikshit Shah – Research Interests

My research focuses on developing optimization-driven methodology for an array of problems in machine learnig, data analysis and their applications to internet-scale problems encountered at Yahoo. In the past, I have also worked extensively on developing computational tools for problems at the intersection of systems theory, statistics, and signal processing. Key to this approach are understanding notions of simplicity in large-scale problems via an appropriate notion of algebraic structure, and using convex optimization as a means to exploit it to develop efficient computational algorithms. In the process I enjoy understanding the conceptual connections between optimization, systems theory, and statistics/applied probability. As an example, see my recent paper with Venkat Chandrasekaran on signomial optimization using relative entropy relaxations, for which we were awarded the INFORMS Optimization Society Young Researcher Prize.