Colloquia Patavina: quarta conferenza


Il 16 gennaio 2007 alle ore 16:00 in aula 1A/150 il Professor Marco Gori dell'Università di Siena terrà (in italiano) la quarta conferenza della serie Colloquia Patavina. Trovate qui sotto un sommario della conferenza ed una breve biografia (in inglese).

Ci auguriamo che parteciperete numerosi. Come al solito, alla conferenza seguira' un piccolo rinfresco.

La Commissione Colloquia (Garuti, Pavon, Pitteri, Rossi).
- Titolo
Learning in web domains

- Conferenziere
Marco Gori
Dipartimento di Ingegneria dell'Informazione
Università di Siena

Martedì 16 Gennaio 2007, ore 16
Aula 1A/150, Torre Archimede
v.Trieste 63

- Abstract
A remarkable number of important problems in different domains (e.g. web mining, pattern recognition, biology...) are naturally modeled by functions defined on graphical domains, rather than on traditional vector spaces. In this talk, I propose the class of Unfolding-Equivalence Functions (UEF), whose computation can be carried out using PageRank-like schemes. Using arguments from function approximation theory, I argue that the class UEF can represent nearly any conceivable Web page rank, including PageRank, HITS, TrustRank, and generally any function defined by human-expert supervision. I sketch the architecture of a† learning machine that calculates any UEF function class by the appropriate selection of a set of parameters. They can be discovered from supervised examples using neural-like optimization algorithms. Interestingly, the analysis of the learning machine benefits from recent advances in the field of calculus for functions defined on graphs.

I give some very promising experimental results on artificial tasks and on the learning of PageRank. Interestingly, similar performance holds also for extensions of PageRank in which the function also depends on the content of the page. Finally, I argue that the propagation of the relationships expressed by the links reduces dramatically the sample complexity with respect to traditional learning machines, thus making it reasonable the application to real-world problems on the Web, like spam detection and complex page categorization.

- Brief bio
Marco Gori received the Ph.D. degree in 1990 from Universita' di Bologna, Italy. From October 1988 to June 1989 he was a visiting student at the School of Computer Science (McGill University, Montreal). In 1992, he became an Associate Professor of Computer Science at Universita' di Firenze and, in November 1995, he joined the Universita' di Siena, where he is currently full professor of computer science.

His main interests are in machine learning, with applications to pattern recognition, Web mining, and game playing. He is especially interested in the formulation of relational machine learning schemes in the continuum setting. He is the leader of the WebCrow project for automatic solving of crosswords, that has recently thrashed human competitors in an official competition taken place within the ECAI-06 conference. He is co-author of the book "Web Dragons: Inside the myths of search engines technologies," Elsevier, 2006.

Dr. Gori serves (has served) as an Associate Editor of a number of technical journals related to his areas of expertise, including IEEE Transaction on Neural Networks, Pattern Recognition, Neural Networks, Neurocomputing, Pattern Analysis and Application, the International Journal of Document Analysis and Recognition, and the International Journal on Pattern Recognition and Artificial Intelligence. He has been the recipient of best paper awards and keynote speakers in a number of international conferences.

He is the Chairman of the Italian Chapter of the IEEE Computational Intelligence Society, a fellow of the ECCAI and of the IEEE.

Rif. M. Pavon