Avrim Blum, Jon Kleinberg, Jianbo Shi, Eva Tardos, Ramin Zabih

Graph cuts and separators of various forms have a long history in Algorithms. More recently, they have been used in Computer Vision for problems of image segmentation and data cleaning, among others. In Machine Learning, there has been increasing interest in problems of learning from labeled and unlabeled data, as well as probabilistic inference when data have pairwise relationships, that seem closely related to notions of graph partitioning. However, in each area, the objectives are subtly different, and it's not always clear how to best formalize them. The purpose of this workshop is to bring together researchers in Algorithms, Vision, and Machine Learning around the subject of Graph Partitioning, in order to discuss and better understand the connections between these problems and the techniques used to solve them. The workshop will be a combination of survey talks, new results and informal discussion. There will be no published proceedings, but we plan to have a web page for the workshop with slides, pointers to relevant papers, and so forth.

The ALADDIN Workshop on Graph Partitioning in Vision and Machine Learning,CMU, January 9-11, 2003