In his History of Western Philosophy, Bertrand Russell claimed that a special section of Hell was reserved for those who claimed to have refuted David Hume on the impossibility of establishing causality. Here (via Isabelle Guyon in e-mail) is your chance to risk damnation in exchange for valuable cash prizes, and a paper in JMLR.
Deadline April 30, 2008
This challenge bridges the gap between data mining/machine learning and causal discovery. Several datasets drawn from real data, or emulating real data, are provided, with the goal of making predictions under "manipulations".
The setting is similar to a usual predictive modeling setting: We have a training set and a test set; a target variable, whose values are concealed in test data, must be predicted. But, the test data are not distributed like the training data: some variables in test data are "manipulated" by an external agent, i.e. set to given values instead of being drawn from the "natural" distribution. Such problems are encountered in many application domains: In medicine to predict the effect of a new treatment, in economy or ecology to predict the consequences of new issued policies, in marketing to predict customer response to marketing campaigns. We anticipate that the tasks of the challenge should require the knowledge of causal relationships between variables since acting on causes of the target may result in a response change while acting on consequences should not. However, we encourage participants to enter the challenge with any approach to the problem.
Despite being nearly synonymous with causality among machine-learners, the use of graphical models is not required — they're serious about the "any approach" bit. (Parochial boosterism, however, leads me to guess that the winner will use graphical models extensively.) If you're interested, do check out the contest homepage, especially the FAQ.
Mr. Hume and Lord Russell could not be reached for comment.
Posted at February 11, 2008 19:48 | permanent link