mwojnars ( on December 2, 2009, 16:20:13 ) I'd like to invite you to RSCTC'2010 Discovery Challenge which started on Dec 1, 2009 at TunedIT, as a special event of the 7th International Conference on Rough Sets and Current Trends in Computing. The task is related to feature selection in analysis of DNA microarray data and classification of patients for the purpose of medical diagnosis and treatment. Prizes worth over 3,000 USD will be awarded to the best solutions. Challenge ends on Feb 28, 2010. This challenge has a unique feature not found elsewhere, related to OSS. Namely, on the Advanced track (there's also Basic), participants submit Java source code of a machine-learning algorithm. The code is compiled on server and automatically evaluated with TunedTester application. Evaluation consists of training the algorithm-classifier on a training subset of data and testing on the test subset. Since we've got the actual code of the algorithm in hand, not only a text file with predictions generated by participant, we can test it very precisely: run multiple train+test trials (like cross-validation), measure execution time, memory requirements etc. Even more, at the end of the challenge, source code of the best algorithms will reside on TunedIT server. If only the authors agree, we will make this code publicly available. Certainly, every participant may publish his code on his own, but here's a guarantee, assured by challenge organizers, that the published code is exactly the code evaluated during the challenge. I invite everyone to participate. Challenge web page: http://tunedit.org/challenge/RSCTC-2010-A Marcin Wojnarski

Post a Reply