Ting Zhang
University of Georgia, GA·Mathematics & Statistics
Many kinds of data are collected over time: stock prices for thousands of companies, brain scans, phone usage across millions of users, power-grid readings and climate records. Analysts want to know two things about such data.
- Fields
Mathematics & Statistics
Computer Science
- Topics
change-point detection
change point analysis
irregular signals
- Grants
- 2 known grants, through Aug 2029
What they work on
- Develop change-point tests that stay powerful when the post-change signal varies unpredictably, using a causal representation framework and Gaussian approximation to set statistically sound alarm thresholds.
- Extend detection of irregular changes from offline analysis to online sequential monitoring, for settings both with and without stable historical data, with theoretical guarantees for dependent, noisy data.
- Characterize analytically how the uncertainty of self-normalized statistics accumulates as the number of variables in a time series grows, using tools from harmonic analysis.
- Build self-normalized methods for high-dimensional time series, including feature selection, simultaneous uncertainty quantification, and inference for medians, quantiles, variances and autocovariances.
More about the research
When does something suddenly change? And how sure can we be about patterns in data with huge numbers of variables, where each measurement depends on earlier ones? Many standard statistical tools assume that the data points are independent of each other, or that a change takes a simple, predictable form. Two projects develop new statistical methods, with mathematical guarantees, for data that break these assumptions.
Why it matters
Reliable alarms for sudden, unpredictable changes could help detect power-grid disruptions, attacks on sensor networks and shifts in social networks. Such alarms could also be harder to fool with deliberately designed adversarial inputs. Valid error estimates for large, time-dependent datasets could support sounder conclusions in economics, finance, neuroscience brain imaging, telecommunications and climate science.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Benjamin Boniece
Drexel University, PA·Mathematics & Statistics
Develop methods to detect shocks that affect many series in high-dimensional, high-frequency data, possibly asynchronously, using time-localized principal component analysis in factor-structured models.
Yao Zheng
University of Connecticut, CT·Mathematics & Statistics·Early-career
Develop recurrent neural network (RNN) models for high-dimensional time series that stay interpretable through Granger causality, a test of whether one series predicts another, to balance accuracy and interpretability.
Yinqiu He
University of Wisconsin-Madison, WI·Mathematics & Statistics
Develop methods with theoretical guarantees for estimating low-dimensional network embeddings and quantifying their uncertainty under general network models with relaxed assumptions.
The school
University of Georgia: 27 labs on PI Pages, $628M a year on research, 598 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.