Location: Geology Building 4660

Xiaotong Shen, Distinguished Professor
College of Liberal Arts, University of Minnesota

Abstract:

High-dimensional generative modeling requires a delicate balance between support fidelity, keeping generated data precisely on the data manifold, and sampling efficiency. While diffusion models capture manifold structures well, their iterative denoising steps are slow and can leak off-support. Conversely, normalizing flows offer rapid, one-pass sampling but are limited by strict invertibility constraints. In this talk, we introduce MAGT (Manifold-Aligned Generative Transport), a novel flow-like generator that achieves the best of both worlds. MAGT learns a one-shot, manifold-aligned transport from a low-dimensional base distribution directly to the data space. By training at a fixed, numerically stable Gaussian smoothing level and approximating the score via latent anchor points, we create a tractable objective without sacrificing speed. We will demonstrate how MAGT samples in a single forward pass, tightly concentrates probability on the learned support, and enables principled likelihood evaluation. We will also share theoretical Wasserstein bounds and empirical results showing significant improvements in both fidelity and sampling speed over standard diffusion models.

This is joint work with Xinyu Tian at the University of Minnesota.

Bio:

Xiaotong T. Shen is the John Black Johnston Distinguished Professor in the College of Liberal Arts at the University of Minnesota. He received his Ph.D. in Statistics from the University of Chicago in 1991.

Professor Shen’s research spans a broad range of modern statistical and machine learning methodologies, including high-dimensional inference, non- and semiparametric methods, causal inference, graphical models, explainable machine intelligence, personalization and recommender systems, natural language processing, generative modeling, and nonconvex optimization. His current work focuses on advancing causal and constrained inference, as well as generative inference and prediction for black-box learning systems, with particular emphasis on diffusion models, normalizing flows, and large-scale summarization. His research is motivated by impactful applications in biomedical sciences, artificial intelligence, and engineering.

Professor Shen has played a significant leadership role in the statistical community through his extensive editorial service. He has served on the editorial boards of leading journals such as the Journal of Machine Learning Research, the Journal of the American Statistical Association, and the Annals of Statistics. He is a former Co-Editor of Statistica Sinica and currently serves as Co-Editor of JASA. His contributions have been widely recognized. He is an elected Fellow of the Institute of Mathematical Statistics, the American Statistical Association, and the American Association for the Advancement of Science. His honors include the “Scholar of the College” award and the ICSA Distinguished Achievement Award.