BEGIN:VCALENDAR VERSION:2.0 PRODID:-//jEvents 2.0 for Joomla//EN CALSCALE:GREGORIAN METHOD:PUBLISH BEGIN:VTIMEZONE TZID:America/Bahia BEGIN:STANDARD DTSTART:20180516T100000 RDATE:20380119T001407 TZOFFSETFROM:-0300 TZOFFSETTO:-0300 TZNAME:America/Bahia -03 END:STANDARD END:VTIMEZONE BEGIN:VEVENT UID:07856eab4cb4d6144f91e3782cf7d9f1 CATEGORIES:Eventos PESC (Palestras, Seminários, etc.) SUMMARY:Seminário: Prof. Bruno Ribeiro (Univ. de Purdue, EUA) DESCRIPTION;ENCODING=QUOTED-PRINTABLE:
Dia 17/05, sexta-feira, teremos mais uma visita do Prof. Bruno Ribeiro ( Univ. de Purdue, EUA), que agora também vem se aventurando no furacão de re des neurais profundas, trazendo para a área sua bagagem em modelagem probab ilística em redes. Sua palestra no PESC irá abordar o problema de represent ação eficiente de grafos para aprendizado, um dos problema quentes desta ár ea no momento (detalhes abaixo). A palestra será abrangente e acessível a u m público não especializado no tema.
Programe-se, participe e ajudem na divulgação! Mais detalhes abaixo ou clicando aqui.
------
Palestrante: Bruno Ribeiro, Assistant Professor, Purdue University, EUA
Título: Graph Re presentation Learning: Where Probability Theory, Data Mining, and Neural Ne tworks Meet
Dia/horário/local: 17/5 (sexta) - 10h - sala H-324B
Resumo:
My talk starts by turning back the clock to 1979-198 3, introducing the ideas that culminated with the fundamental representatio n theorem of graphs (the Aldous-Hoover theorem). I will then show how these ideas connect to a probabilistic interpretation of matrix factorization me thods, explaining why matrix factorization is fundamentally not as expressi ve as it could be to describe finite graphs. I will then turn to early mach ine learning attempts to represent graphs and how these attempts connect to graph mining algorithms. I will introduce the concept of representation le arning with graph neural networks (GNNs) and explain its connections to sta tistical graph models and the Weisfeiler-Lehman isomorphism test. Finally, I will introduce a newly proposed general framework for graph representatio n learning using deep neural networks, which is directly rooted in the idea s that gave us the Aldous-Hoover representation theorem. This new represent ation framework points to novel graph models, new approaches to make existi ng methods scalable, and provides a unifying approach connecting matrix fac torization, graph mining algorithms, and graph neural networks. I will end my talk with a few open problems.
This talk is in part based on joint work with Ryan Murphy, Balasubramanian Srinivasan, and Vinayak Rao.
Bio resumida:
Bruno Ribeiro is an Assistant Professor in the Department of Computer Science at Purdue University. He obtained his Ph.D. at the University of Massachusetts Amherst and did his postdoctoral studies at Carnegie Mellon University from 2013-2015. His research interests are i n deep learning and data mining, with a focus on sampling and modeling rela tional and temporal data.
DTSTAMP:20240328T092231Z DTSTART;TZID=America/Bahia:20190517T100000 DTEND;TZID=America/Bahia:20190517T120000 SEQUENCE:0 TRANSP:OPAQUE END:VEVENT END:VCALENDAR