CNRS and University of Paris South, France
PASCAL (Pattern Analysis, Statistical Learning and Computational Learning) Network of Excellence had a Challenge programme between 2003 and 2013. The circa 50 funded challenges aimed at i) advancing the state of the art in the fields of computer vision and natural language processing; ii) exploring new ML trends, and iii) pioneering new fields of application.
The talk will discuss a few issues behind challenge design (e.g. large scale manual labeling, assessing computational complexity, designing measures of assessment, dealing with pluri-disciplinarity). The opportunities and pitfalls related to the organization of a series of challenges on a given theme will also be discussed.