Fall 2026 Seminar Series
Department of Statistics and Data Science 2026-2027 Seminar Series - Fall 2026
The 2026-2027 Seminar Series will primarily be in person, but some talks will be offered virtually using Zoom. Talks that are virtual will be clearly designated and registration for the Zoom talks will be required to receive the zoom link for the event. Please email Kisa Kowal at k-kowal@northwestern.edu if you have questions.
Seminar Series talks are free and open to faculty, graduate students, and advanced undergraduate students
From A Cauchy Surprise to The Half-Cauchy Miracle
Friday, October 23, 2026
Time: 11:00 a.m. to 12:00 p.m. central time
Location: Ruan Conference Room – lower level (Chambers Hall 600 Foster Street)
Speaker: Xiao-Li Meng, Whipple V. N. Jones Professor of Statistics, Harvard University
Abstract: This talk follows the path from Pillai and Meng (2016, Annals of Statistics), “ An unexpected encounter with Cauchy and Levy” to Liu, Meng, and Pillai (2025, https://arxiv.org/abs/2501.01065) “A Heavily Right Strategy for Integrating Dependent Studies in Any Dimension”, inviting the audiences to join a journey to explore an emerging and mystic force in principled statistical inference in arbitrarily high dimensions: heavy-tail approximations.
This talk will be given in person on Northwestern's Evanston campus.
planitpurple.northwestern.edu/event/647570
Categorical distance correlation: Concepts, properties and applications
Friday, November 6, 2026
Time: 11:00 a.m. to 12:00 p.m. central time
Location: Ruan Conference Room – lower level (Chambers Hall 600 Foster Street)
Speaker: Qingyang Zhang, Associate Professor of Mathematical Sciences, University of Arkansas
Abstract: In this talk, I will introduce categorical distance correlation (CDC), a simple yet powerful statistical functional for assessing dependence in categorical data. After highlighting its empirical power advantages over the traditional Chi-squared test, I will present key theoretical properties of CDC. These include its B-robustness for fixed or diverging numbers of categories, its asymptotic distributions under both null and alternative hypotheses, and the sure screening properties of the maximum likelihood and unbiased estimators. I will also demonstrate its practical utility through an application to General Social Survey data. Time permitting, I will discuss two extensions, including CDC under general encodings (such as one-hot encoding for nominal variables and semi-circle encoding for ordinal variables) and a privacy-preserving framework for CDC.
This talk will be given in person on Northwestern's Evanston campus.
planitpurple.northwestern.edu/event/647057
Counterfactual JEPAs for World Models in the Real World
Friday, November 13, 2026
Time: 11:00 a.m. to 12:00 p.m. central time
Location: Ruan Conference Room – lower level (Chambers Hall 600 Foster Street)
Speaker: Randall Balestriero, Assistant Professor of Computer Science, Brown University; researcher at AMI Labs
Abstract: This talk presents a sequence of recent results on Joint-Embedding Predictive Architectures (JEPAs), spanning self-supervised representation learning, action-conditioned world models, and counterfactual reasoning. We begin with LeJEPA, which addresses representation collapse through a distributional principle: embeddings are regularized toward an isotropic Gaussian using Sketched Isotropic Gaussian Regularization, producing a simple predictive objective with theoretical guarantees and without teacher–student networks or stop-gradient heuristics. We then introduce LeWorldModel, which extends this construction to sequential data by learning an encoder and action-conditioned latent transition model, (z_{t+1} \approx f_\theta(z_t,a_t)), end-to-end from raw observations. By combining next-embedding prediction with Gaussian latent regularization, the model supports planning and control directly in representation space, without reconstructing future pixels. Finally, we discuss new directions in counterfactual JEPAs, where object-level masking and latent interventions create structured partial-observability problems that force the predictor to model interaction-dependent dynamics rather than exploit local correlations. Throughout, we emphasize the geometric and statistical principles connecting these methods, and ask when latent prediction is sufficient for learning causal, controllable abstractions of a dynamical world.
This talk will be given in person on Northwestern's Evanston campus.
planitpurple.northwestern.edu/event/647674