Assistant Professor of Statistics
Yao Zheng completed her BS and Ph.D. in the Department of Statistics and Actuarial Science at the University of Hong Kong in 2013 and 2017, respectively. She previously studied in the School of Economics and Management at Tsinghua University from 2008 to 2009 and in the Department of Mathematics at UCLA in 2011. Before joining the Department of Statistics at UConn, she was a visiting assistant professor and a postdoctoral fellow at Purdue University’s Department of Statistics.
Q&A with Yao Zheng
What are your research interests?
My research interests lie at the intersection of statistics, machine learning, econometrics, and optimization. Specifically, I specialize in statistical learning of time-dependent data. My current research focuses on two main problems: (1) non-asymptotic statistical learning and (2) statistical and computational methods for the analysis of many time series. The objective of the former is to develop statistical methods that work well even when the sample size is very small, which usually cannot be achieved by conventional asymptotic tools. The objective of the latter is to develop an efficient, robust, and computationally feasible statistical inference procedure that accommodates the high dimensionality of the data.
How did you become interested in this type of work?
I was first drawn to time series analysis because of the unique and interesting modeling techniques and theoretical challenges of temporal data. For my recent work on non-asymptotic theory, I was motivated by the fact that, unlike time series analysis in classical statistics, theoretical properties of stochastic processes in machine learning are commonly studied via non-asymptotic tools. I thought it would be interesting to borrow insights and practical motivations from machine learning to develop new advanced statistical tools. So far, most full-fledged statistical learning theory has been confined to independent data, whereas data are collected over time in many modern applications, and there is a need to analyze data in an on-line (sequential) setting. This motivated me to explore methods that can efficiently and robustly incorporate the temporal structure as well as the possible high dimensionality in various learning tasks.
What drew you to UConn?
First, the Storrs campus has a wonderful location. It is peaceful, beautiful, and convenient. I like working and living in such an environment. Second, the research interests of the faculty in the Department of Statistics span virtually all major statistical specializations, and I have thoroughly enjoyed my interactions with the faculty members. Moreover, the Department has an open and harmonious academic environment, which I believe will be very helpful for my research and teaching career.