The project supports graduate student participation in the 28th International Conference on Machine Learning (ICML 2011). Specifically, the project supports travel to the conference for those who might not otherwise be able to attend for financial reasons, and organizes a student poster-presentation program that will facilitate one-on-one discussions and other mentoring with the world's leading researchers in machine learning. Students are exposed to state-of-the-art work by other researchers and have the opportunity to attend tutorials on material that is not taught at their home institutions. Participating students receive feedback from senior researchers beyond their institutional and national boundaries. Furthermore, participation in the poster session and conference helps to integrate these students into the research community and represents a natural integration of research and education.

Project Report

NSF's generous $20,000 grant for student travel was enormously helpful for enabling student travel to the conference in Seattle. Using the NSF funds, ICML was able to support a total of 26 of 44 student applicants, The technical program for ICML-2010 was very strong and it gave students (and other attendees) had a wonderful chance meet people, become energized, and to learn about the field. ICML was, as always, quite selective in its choice of papers, but the technical program still contained no less than 170 contributed papers, each of which was presented both orally and in a poster session. There were also four excellent invited talks, from Christopher Bishop, Harmut Neven, Martin Nowak and David Ferrucci. Lessons Learned Overall, the process went relatively smoothly. Hal Daume from UMD was in charge for fund rasing, and Jesse Davis from Kuleuven was in charge of overall student award and selection process. To the extent possible, we tried to adapt to the lessons learned from previous years, but were not entirely successful, so some of them are repeated below. A better collection process of all required information from students beforehand. Students filled out an online form, but later we found that we needed additional information (social security numbers). All this should be collected at one or at most two time points, as early on in the process as possible. More timely reimbursements and better communication of when to expect the reimbursements. These were both issues from the previous year and continued to be a problem for us. Additionally, we made an attempt this year to reach out to underrepresented groups and to underrepresented universities. This involved both an initial push by faculty to tell their colleagues about this specific goal, a flurry of emails to faculty at underrepresented universities (ones that do not typically send students to ICML or ones in EPSCORE states), and a specific note on the call for applications. It is difficult to measure the success rate here. In previous years, ICML was unable to support any students from underrepresented universities simply because they did not apply. This year, we received applications from four such universities (University of Iowa, University of Wyoming, Utah State University and Depaul University). All four of these students were supported with a total of $4100 from NSF. Out of the US pool, we additionally received applications from only five female students (out of 26, with two who did not specify gender), all of which were supported (though two of whom also were from underrepresented universities). For comparison, the male/female ratio was essentially the same for non-US applicants. While this is certainly better than last year (when there were no applicants from underrepresented universities), we feel that we can still do a better job getting the word out to students. Of all 53 applicants, all but four had accepted papers at ICML (those four are precisely the four from underrepresented universities). This suggests that, despite our efforts, students who do not have papers at ICML are not applying for support to attend ICML. We need to do a better job getting word out that it is encouraged for underrepresented students to apply, even if they don't have a paper (and perhaps especially if they don't have a paper). Costs and Accounting The 2011 scholarship program received applications from 53 students. We offered scholarships to 44 students (22 from the US), ranging in travel funds from $400-$1830 for US students depending on needs and distance to Seattle. The total funding given to support students in 2008 was $33,800. The program was funded through NSF and through contributions from private sources. Last year all students who had accepted papers at the conference or associated workshops received support. Information provided by the students who were supported by the NSF funding (23 of the 25 US students were funded by the NSF) is included at the end of the report. Summary In summary, the Student Poster and Travel Program for the 2011 International Conference on Machine Learning was extremely successful and enabled a number of students to attend the conference who otherwise might not have been able to attend. Moreover, the design of the program insured that students received valuable feedback on their research. Those of us on the Board of the International Machine Learning Society (IMLS) that organizes the ICML Conference are grateful for the continued support from NSF that has (and continues to) help many young researchers and improve the quality of the annual ICML meeting.

Project Start
Project End
Budget Start
2011-04-01
Budget End
2012-03-31
Support Year
Fiscal Year
2011
Total Cost
$20,000
Indirect Cost
Name
University of Maryland College Park
Department
Type
DUNS #
City
College Park
State
MD
Country
United States
Zip Code
20742