Case Studies in Bayesian Statistics and Machine Learning continues the tradition of the workshop series Case Studies in Bayesian Statistics with a meeting October 14-15, 2011. The usual format for meetings and workshops in statistics and computer science emphasize methods over applications, which often stifles discussion about the impact of the methods on the substantive problem. The unique format of the case studies workshop series has allowed for substantive discussion of application-specific issues, most importantly a narrative of how the substantive scientific problem demanded either new methods or the novel application of existing approaches, the obstacles in the real problem that the researchers encountered, and the solutions that resulted.
Statistics and machine learning provide essential methodologies throughout the sciences, yet the connection between the science and the data analytic problem formulation is rarely emphasized in traditional conferences. This workshop fosters cross-pollination of ideas from the statistics and computer science ommunities and promotes work that takes on important challenges in scientific investigation. The case studies highlight the way novel application of statistical machine learning methods are used to answer a scientific question. The workshop especially supports efforts by young investigators, in part by including a session of presentations exclusively by younger investigators.
workshop was held October 14th and 15th, 2011. Four invited case studies presentations were given on topics in proteomics, forestry, brain imaging, and cosmology. The workshop highlighted the application of advanced Bayesian and statistical learning methods to difficult scientific problems in these areas. The originality of the case studies format, compared to other scientific conferences in statistics and machine learning, stems from the balance in the presentation between the science and the statistical modeling. The second key component of the workshop was a session devoted to short presentations from young researchers on their applied work. Five junior researchers presented their work on problems in telecommunications, computing systems, high energy physics, public health and sexuality, and systems biology. The third component of the workshop was the DeGroot Lecture, given by Daphne Koller, professor of Computer Science at Stanford University. Professor Koller is a world-renowned expert in probabilistic learning and computer vision. The fourth component to the workshop was the evening banquet and poster session, with most contributions coming from graduate students and junior researchers. The workshop attracted 110 participants, a large fraction of whom were graduate students and junior researchers. The workshop very successfully emphasized the scientific problems motivating the statistical and machine learning methods. In fact, every case study talk had one speaker from the scientific area who discussed the scientific field and one speaker who specialized in the statistical methods. The junior researcher presentation session and the poster session gave young researchers ample opportunity to present their research to senior colleagues. During coffee breaks, the poster session and the banquet, students were able to closely mingle with senior researchers. We also were very successful in attracting women to the workshop, including the keynote speaker, one of the invited case studies (presented by two women), two of the young researcher presentations and several posters. One of the most important contributions of this workshop was the interaction between Bayesian researchers (traditionally from statistics and biostatistics departments) and machine learning researchers. Several workshop attendees commented on the heterogeneity of the talks and the posters and indicated that the workshop provided a unique experience.