{"688808":{"#nid":"688808","#data":{"type":"event","title":"PhD Defense by Xinyuan Cao","body":[{"value":"\u003Cp\u003E\u003Cstrong\u003ETitle:\u0026nbsp;\u003C\/strong\u003EFoundations of Efficient Representation Learning\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EDate:\u0026nbsp;\u003C\/strong\u003EMarch 17, 2026\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ETime:\u0026nbsp;\u003C\/strong\u003E12:00 pm - 2:00 pm ET\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ELocation\u003C\/strong\u003E: Klaus 1212\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EZoom link\u003C\/strong\u003E: \u003Ca href=\u0022https:\/\/gatech.zoom.us\/j\/94992205747\u0022 title=\u0022https:\/\/gatech.zoom.us\/j\/94992205747\u0022\u003Ehttps:\/\/gatech.zoom.us\/j\/94992205747\u003C\/a\u003E\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EXinyuan Cao\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003EMachine Learning PhD Student\u003C\/p\u003E\u003Cp\u003ESchool of Computer Science\u003Cbr\u003EGeorgia Institute of Technology\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ECommittee\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003E1 Dr. Santosh Vempala\u0026nbsp;(Advisor), School of Computer Science, Georgia Institute of Technology\u003C\/p\u003E\u003Cp\u003E2 Dr. Jacob Abernethy, School of Computer Science, Georgia Institute of Technology\u003C\/p\u003E\u003Cp\u003E3 Dr. Pan Li, School of Electrical and Computer Engineering, Georgia Institute of Technology\u003C\/p\u003E\u003Cp\u003E4 Dr. Sahil Singla, School of Computer Science, Georgia Institute of Technology\u003C\/p\u003E\u003Cp\u003E5 Dr. Freda Shi, School of Computer Science, University of Waterloo\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EAbstract\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003ERepresentation learning extracts lower-dimensional, structured features from complex, unstructured data and reuses them across tasks. Despite strong empirical performance, its theoretical foundations remain limited. This thesis bridges this gap by developing formal efficiency guarantees for representation learning.\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003EFirst, we study unsupervised identification of geometric structure and give a polynomial-time algorithm that recovers a halfspace with margin from unlabeled data under broad distributional conditions. Next, we analyze implicit structure in sequence modeling. By formalizing long-range structure using efficient distinguishers, we prove that minimizing next-token prediction loss over bounded-size networks yields an indistinguishable language model, with model size polynomial in the distinguisher parameters and independent of document length. Finally, we study how learned structures can be efficiently transferred in lifelong learning, where tasks arrive sequentially and the model continually refines the representation while maintaining performance on earlier tasks. We propose algorithms that dynamically learn, refine, and reuse features across sequential tasks, achieving near-optimal sample complexity.\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003ETogether, these results provide a unified learning-theoretic foundation for efficient representation learning, spanning how structure can be identified, induced by training objectives, and transferred across tasks.\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E","summary":"","format":"limited_html"}],"field_subtitle":"","field_summary":[{"value":"\u003Cp\u003EFoundations of Efficient Representation Learning\u003C\/p\u003E","format":"limited_html"}],"field_summary_sentence":[{"value":"Foundations of Efficient Representation Learning"}],"uid":"27707","created_gmt":"2026-03-09 14:13:39","changed_gmt":"2026-03-09 14:14:24","author":"Tatianna Richardson","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2026-03-17T12:00:00-04:00","event_time_end":"2026-03-17T14:00:00-04:00","event_time_end_last":"2026-03-17T14:00:00-04:00","gmt_time_start":"2026-03-17 16:00:00","gmt_time_end":"2026-03-17 18:00:00","gmt_time_end_last":"2026-03-17 18:00:00","rrule":null,"timezone":"America\/New_York"},"location":"Klaus 1212","extras":[],"groups":[{"id":"221981","name":"Graduate Studies"}],"categories":[],"keywords":[{"id":"100811","name":"Phd Defense"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[{"id":"1788","name":"Other\/Miscellaneous"}],"invited_audience":[{"id":"78771","name":"Public"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}