Search for a command to run...
This volume contains the papers accepted to the 25th International Conference on Machine Learning (ICML 2008). ICML is the annual conference of the International Machine Learning Society (IMLS), and provides a venue for the presentation and discussion of current research in the field of machine learning. These proceedings can also be found online at http://www.machinelearning.org. This year, ICML was held July 5..9 at the University of Helsinki, in Helsinki, Finland, and was co-located with COLT-2008, the 21st Annual Conference on Computational Learning Theory, and UAI-2008, the 24th Conference on Uncertainty in Artificial Intelligence. No less than 583 papers were submitted to ICML 2008. There was a very thorough review process, in which each paper was reviewed double-blind by three program committee (PC) members. Authors were able to respond to the initial reviews, and the PC members could then modify their reviews based on online discussions and the content of this author response. There were two discussion periods led by the senior program committee (SPC), one just before and one after the submission of author responses. At the end of the second discussion period, the SPC members gave their recommendations and provided a summary review for each of their papers. Some papers were checked by the SPCs to ensure that reviewer comments had been addressed. Apart from the length restrictions on papers and the compressed time frame, the review process for ICML resembles that of many journal publications. In total, 158 papers were accepted to ICML this year, including a small number of papers which were initially conditionally accepted, yielding an overall acceptance rate of 27%. ICML authors presented their papers both orally and in a poster session, allowing time for detailed discussions with any interested attendees of the conference. Each day of the main conference included one or two invited talks by a prominent researcher. We were very fortunate to be able to host Michael Collins, of the Massachusetts Institute of Technology; Andrew Ng, of Stanford University; and Luc De Raedt, of the Katholieke Universiteit Leuven, and John Winn of Microsoft Research Cambridge. In addition to the technical talks, ICML- 2008 also included nine tutorials held before the main conference, presented by Alex Smola, Arthur Gretton, and Kenji Fukumizu; Bert Kappen and Marc Toussaint; Neil Lawrence; MartinWainwright; Ralf Herbrich and Thore Graepel; Andreas Krause and Carlos Guestrin; Shai Shalev-Shwartz and Yoram Singer; Rob Fergus; and Matthias Seeger. This year our workshops were organized jointly with COLT and UAI as part of a special overlap day, consisting of eleven workshops selected and arranged collaboratively by the respective workshop chairs of the three conferences. This day provided a rich opportunity for interaction among the attendees of the conferences. This year, ICML enlarged its award offerings to match several other well-established conferences. We hope these will help build our community, celebrate our advances, and encourage applications and long-term thinking. In addition to our previously traditional Paper and Student Paper awards, we also gave awards for Application Paper and 10-year Best Paper (for the best paper of ICML 1998, optionally given in conjunction with a co-located conference). We thank the Machine Learning Journal for sponsoring some of our paper awards.