feature selection in analysis of DNA microarray data and classification of patients for the purpose of medical diagnosis and treatment
RSCTC'2010 Discovery Challenge
is a special event of the 7th International Conference on Rough Sets and Current Trends in Computing (www.rsctc2010.org) that will take place in Warsaw, Poland, June 28-30, 2010. The competition is organized by
TunedIT (tunedit.org) in close co-operation with organizers of the conference. 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.
In recent years, a lot of attention of researchers from many fields has been put into investigation of DNA microarray data. This growing interest is largely motivated by numerous practical applications of knowledge acquired from such data in medical diagnostics, treatment planning, drugs development and many more. When analyzing microarray data, researchers have to face the few-objects-many-attributes problem, as the usual ratio between the number of examined genes and the number of available samples exceeds 100. Many standard classification algorithms have difficulties in handling such highly dimensional data and due to low number of training samples tend to overfit. Moreover, usually only a small subset of examined genes is relevant in the context of a given task. For these reasons, feature extraction methods - in particular the ones based on rough-set theory and reducts - are an inevitable part of any successful microarray data classification algorithm. With RSCTC'2010 Discovery Challenge we would like to stimulate investigation in these important fields of research.
The challenge comprises two independent tracks, Basic and Advanced, differing in the form of solutions. In Basic track, the participant submits a text file with predicted decisions for test samples, which are then compared with ground truth decisions. In Advanced track, the participant submits Java source code of a classification algorithm, which is compiled and automatically evaluated on the server. Although Advanced track is more challenging for participants than Basic one, it enables much more precise evaluation of solutions and encourages sharing of algorithm implementations with the community after the end of the contest.
Challenge web page: tunedit.org/challenge/RSCTC-2010-A
Start: Dec 1, 2009 - already started!
End: Feb 28, 2010