BEGIN:VCALENDAR VERSION:2.0 PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4// BEGIN:VEVENT UID:20260416T024441EDT-3873euE3Li@132.216.98.100 DTSTAMP:20260416T064441Z DESCRIPTION:Title: Regularized Fine-Tuning for Representation Multi-Task Le arning: Adaptivity\, Minimaxity\, and Robustness\n\n \n\nAbstract:\n\nWe s tudy multi-task linear regression for a collection of tasks that share a l atent\, low-dimensional structure. Each task’s regression vector belongs t o a subspace whose dimension\, denoted intrinsic dimension\, is much small er than the ambient dimension. Unlike classical analyses that assume an id entical subspace for every task\, we allow each task’s subspace to drift f rom a single reference subspace by a controllable similarity radius\, and we permit an unknown fraction of tasks to be outliers that violate the sha red-structure assumption altogether. Our contributions are threefold. Firs t\, adaptivity: we design a penalized empirical-risk algorithm and a spect ral method.  Both algorithms automatically adjust to the unknown similarit y radius and to the proportion of outliers. Second\, minimaxity: we prove information-theoretic lower bounds on the best achievable prediction risk over this problem class and show that both algorithms attain these bounds up to constant factors\; when no outliers are present\, the spectral metho d is exactly minimax-optimal. Third\, robustness: for every choice of simi larity radius and outlier proportion\, the proposed estimators never incur larger expected prediction error than independent single-task regression\ , while delivering strict improvements whenever tasks are even moderately similar and outliers are sparse. Additionally\, we introduce a thresholdin g algorithm to adapt to an unknown intrinsic dimension. We conduct extensi ve numerical experiments to validate our theoretical findings.\n\nSpeaker \n\nYang Feng is a Professor of Biostatistics in the School of Global Publ ic Health at New York University\, where he is also affiliated with the Ce nter for Data Science. He earned his Ph.D. in Operations Research from Pri nceton University in 2010. His research centers on the theoretical and met hodological foundations of machine learning\, high-dimensional statistics\ , network models\, and nonparametric statistics\, with applications in Alz heimer’s disease prognosis\, cancer subtype classification\, genomics\, el ectronic health records\, and biomedical imaging\, enabling more accurate models for risk assessment and clinical decision-making. His work has been supported by grants from the National Institutes of Health and the Nation al Science Foundation (NSF)\, including the NSF CAREER Award. He currently serves as Associate Editor for several leading journals\, including the J ournal of the American Statistical Association (JASA)\, the Journal of Bus iness & Economic Statistics\, the Journal of Computational & Graphical Sta tistics\, and the Annals of Applied Statistics. In addition\, he will serv e as Review Editor for JASA and The American Statistician from 2026 to 202 8. His professional recognitions include being named a Fellow of the Ameri can Statistical Association and the Institute of Mathematical Statistics\, as well as an elected member of the International Statistical Institute. \n\nThe presentation will also be accessible online using the following Zo om link\n\nTime: Oct 24\, 2025 03:30 PM Eastern Time (US and Canada)\n\nJo in Zoom Meeting\n\nhttps://mcgill.zoom.us/j/81872329544\n\nMeeting ID: 818 7232 9544\n\n \n DTSTART:20251024T193000Z DTEND:20251024T203000Z LOCATION:Room 1104\, Burnside Hall\, CA\, QC\, Montreal\, H3A 0B9\, 805 rue Sherbrooke Ouest SUMMARY:Yang Feng (New York University) URL:https://www.mcgill.ca/mathstat/channels/event/yang-feng-new-york-univer sity-368448 END:VEVENT END:VCALENDAR