Working Groups of Statistical Interest

Geometric Data Analysis Group

Reading and discussing papers and presenting original work at the intersection of high dimensional geometry, statistics, and machine learning. Currently, the topics of interest are: Topological Data analysis -- what and how? Manifold learning algorithms, estimating the intrinsic dimension, statistical guarantees in geometric learning.

Contacts: Marina Meila & Yen-Chi Chen

Statistical and Computational Optimal Transport

The working group rejuvenates each year around a special topic related to statistical and computational optimal transport. Previous topics include self-attention, transformer architectures, normalizing flows, Schrödinger bridge, and entropy regularized optimal transport.  Participants are expected to have completed the STAT512/STAT513 and STAT581/STAT582/STAT583 sequences.

Contacts: Zaid Harchaoui & Soumik Pal

Statistical and Machine Learning Approaches for the Social Sciences

This group brings together students and faculty inside and outside the UW interested in developing statistical and machine learning methodologies to address open problems in the social sciences. Regular meeting time is typically Wednesday 9am.

Contact: Elena Erosheva

Space-Time Reading Group

The Space-Time Reading Group is a student-run reading group that meets weekly to discuss current topics in spatial and spatiotemporal statistics. We present papers, book chapters, or software under a theme relevant to modelling spatial phenomena. Members include Statistics graduate students as well as professors, researchers and students from across and beyond UW.

Contact: Jon Wakefield

SUCIA-REX: Synthesizing Uncertainty and Complex decisions In the Analysis of vital rates, Relational data, and EXperiments

Our group works on various topics at the interface of statistical methodology and substantive questions in the social and health sciences.  We currently have an emphasis on (i) measuring social networks and understanding the processes that operate through them (using both experiments and observational data), (ii) cost-effective but rigorous data collection for global health and health policy, and (iii) adapting tools from statistics and economics to facilitate decision-making in the presence of predictions from complex statistical learning/artificial intelligence models.  We value deep collaborations with colleagues trained in areas beyond statistics, unexpected connections, and creative acronyms.  We are always excited to hear fresh ideas and perspectives, so please consider dropping by anytime if you're interested.  More info can be found here: thmccormick.github.io.    

Contact: Tyler McCormick

TGIF Lab (Thank Goodness for Influence Functions)

This group consists of students, postdocs, and faculty working on problems in semiparametrics and causal machine learning. Examples of such problems include individualizing treatment recommendations, fusing data from multiple sources, and identifying the immune responses that best predict a vaccine's efficacy. Most meetings consist of presentations by students affiliated with TGIF faculty, though sometimes outside speakers also present. The group meets on Fridays, with the time varying by quarter.

Contacts: Alex Luedtke, Marco Carone, Andrea Rotnitzky, Peter Gilbert