Jasper Lee 李雋軒

  • Fellowship in 2021 at University of Wisconsin-Madison
Dr Jasper Lee is a postdoctoral research associate at the Department of Computer Sciences at UW Madison, hosted by Ilias Diakonikolas. His research focuses on the design of practical, data-efficient algorithms for statistical applications, with rigorous theoretical analysis and guarantees.

Prior to joining UW Madison, Jasper completed his doctoral studies at Brown University, advised by Paul Valiant, and his undergraduate studies at Churchill College, University of Cambridge.

Research Highlight

Consider the following fundamental task in statistics: suppose we are given a set of independent samples drawn from a probability distribution of real numbers, what is the most accurate way to estimate the distribution mean with high confidence? Perhaps surprisingly, it has been known that the most commonly used method, namely computing the average of the samples as the estimate, has far-from-optimal accuracy.

Recent work by Jasper Lee and Paul Valiant resolves this foundational question, under the minimal assumption that the distribution has finite variance. They propose a delicately designed yet easy-to-compute estimator, which they also prove to have the fastest convergence possible. Part of Jasper's postdoctoral research will focus on extending and leveraging this result for further statistical applications.