BEGIN:VCALENDAR PRODID:-//eluceo/ical//2.0/EN VERSION:2.0 CALSCALE:GREGORIAN BEGIN:VEVENT UID:d064b25137e2d46e2fb5485e3d2fa6b0 DTSTAMP:20260608T202503Z SUMMARY:LIDAM Statistics Seminar by Kris Sankaran DESCRIPTION:12/06/2026 - 14:30 - ISBA C.115 - \; \;Kris Sankaran&nb sp\;(University of Wisconsin - Madison) \;Will give a presentation on : \;New Diagnostics for Dimensionality Reduction of Genomic Data   \;Abstract: \;Dimensionality reduction helps organize high-dimensional genomics data into manageable low-dimensional representations\, like diff erentiation trajectories in single cell data and community profiles in met agenomic sequencing. Such reductions are powerful but sensitive to hyperpa rameters and prone to misinterpretation. This talk addresses two risks in dimensionality reduction. First\, we consider how to choose the number of topics K in topic modeling\, a method from population genetics and languag e modeling\, now common in microbiome analysis. While broadly useful\, top ic models require users to specify the number of topics K\, which governs the resolution of learned topics. We discuss a new technique\, topic align ment\, for comparing topics across models with different resolutions. Simu lation studies show that this approach distinguishes between true and spur ious topics. Second\, we examine the distortions introduced by nonlinear d imensionality reduction methods\, like t-SNE and UMAP\, when applied to si ngle cell data. For example\, these methods can introduce spurious cluster s and fail to preserve cell density. We adopt the RMetric algorithm from m anifold learning to measure local distortions. We also develop visualizati ons to explore these distortions. Representing cells as deformed ellipses highlights changes in local geometry\, and an interactive interface select ively reveals evidence for distortion without overwhelming the viewer. Thr ough case studies on simulated and real data\, we find that the visualizat ions can flag fragmented neighborhoods\, support hyperparameter tuning\, a nd enable method selection. Topic alignment and distortion visualization a re available as software packages\, with case studies in online vignettes: https://go.wisc.edu/7h58r9\, https://go.wisc.edu/ss5ts9. URL:https://www.uclouvain.be/en/calendar/isba DTSTART;TZID=Europe/Brussels:20260612T143000 DTEND;TZID=Europe/Brussels:20260612T153000 LOCATION:ISBA - C115 (1st Floor) 1348 Louvain-la-Neuve END:VEVENT BEGIN:VEVENT UID:70d49a3353de05c284b99d4fe4a1d70e DTSTAMP:20260608T202503Z SUMMARY:Applied Statistics Workshop by Kris Sankaran DESCRIPTION:19/06/2026 - 14:30 - ISBA C.115 - \; \;Kris Sankaran&nb sp\;(University of Wisconsin - Madison) \;Will give a presentation on : \;A Brief Introduction to Shapley Values  \;Abstract: \;This workshop will introduce Shapley values\, a game-theoretic concept adapted to explain predictions from black-box machine learning models. We study t he method from statistical and geometric perspectives\, then apply languag e from causal inference to discuss subtleties in its interpretation. Since naive implementations become computationally intractable as samples or fe atures grow\, we review approximation methods like IME and KernelSHAP. We close with practical implementations in R (shapviz) and Python (shap)\, mo tivated by interpretable machine learning problems from social science and developmental biology. Material will be accessible to researchers with ex perience in linear regression and supervised machine learning. The lecture is based off notes for a recently developed undergraduate course on inter pretable machine learning (https://github.com/krisrs1128/stat479_notes).&n bsp\; \; URL:https://www.uclouvain.be/en/calendar/isba DTSTART;TZID=Europe/Brussels:20260619T143000 DTEND;TZID=Europe/Brussels:20260619T143000 LOCATION:ISBA - C115 (1st Floor) 1348 Louvain-la-Neuve END:VEVENT END:VCALENDAR