PASCAL Agnostic Learning
Prior Knowledge

The challenge is now over. But it remains open for post-challenge submissions!

IMPORTANT: Entries made since February 1st 2007 might be using validation data, now available for training.

"When everything fails, ask for additional domain knowledge"

is the current motto of machine learning. Therefore, assessing the real added value of prior/domain knowledge is a both deep and practical question. Most commercial data mining programs accept data pre-formatted as a table, each example being encoded as a fixed set of features. Is it worth spending time engineering elaborate features incorporating domain knowledge and/or designing ad hoc algorithms? Or else, can off-the-shelf programs working on simple features encoding the raw data without much domain knowledge put out-of-business skilled data analysts?

In this challenge, the participants are allowed to compete in two tracks:

Final results of August 1st, 2007

"Prior Knowledge" winner

Vladimir Nikulin

"Agnostic Learning" winner

Roman Lutz
LogitBoost with trees

Individual dataset winners

Prior Knowledge
ADAMarc Boulle with Data Grid (CMA)
GINAVladimir Nikulin with vn2
HIVAChloe Azencott with SVM
NOVAJorge Sueiras with Boost mix
SYLVARoman Lutz with Doubleboost
Agnostic Learning
ADARoman Lutz with LogitBoost with trees
GINARoman Lutz with LogitBoost with trees
HIVAVojtech Franc with RBF SVM
NOVAMehreen Saeed with Submit E final
SYLVARoman Lutz with LogitBoost with trees


Valid challenge entries (*)316
(*) Valid entries include all results for all datasets