Ankit Pensia, recent Computer Sciences PhD graduate, has been awarded the prestigious Herman Goldstine Postdoctoral Fellowship at IBM Research. These fellowships support recent PhD graduates by giving them complete academic freedom to conduct postdoctoral research on problems of their choice in one of the IBM Research laboratories. Pensia says “I am using this opportunity to continue to study statistical inference through the lens of robustness, while also exploring new connections with privacy.”
Pensia’s dissertation was entitled, “Efficient Statistical Inference Under Sampling and Computational Constraints,” and he was advised by UW-Madison faculty members Po-Ling Loh, Varun Jog, and Ilias Diakonikolas. His PhD research focused on algorithms for statistical inference tasks under various statistical and computational constraints. This work broadly lies in the field of robust statistics, where the goal is to develop algorithms that make meaningful predictions even when training data has different kinds of outliers, which are becoming increasingly common in large-scale datasets collected with little to no curation.
As he moves forward, Pensia says, “I enjoy working at the intersection of algorithms and statistics, and there is an abundance of real-world problems that require these tools.” He considers this to be an exciting field to be in, “where one gets to develop mathematically beautiful solutions for meaningful tasks.” Pensia plans to continue working on these problems and forging new collaborations in the field.