(1 - 18 von 19
)
Preference Learning from Qualitative Partial Derivatives
www.ke.tu-darmstadt.de
Jure Zabkar, Martin Moˇ zina, Tadej Janeˇ z, Ivan Bratko, and Janez Demˇ ˇsar University of Ljubljana, Faculty of Computer and Information Science Trˇza ˇska 25, SI Ljubljana, Slovenia -lj.si Abstract. We address the problem of learning preference models from data con-taining implicit preference information.
Rectifying Predictions of Classiflers by Local Rules
www.ke.tu-darmstadt.de
Martin Mo•zina and Ivan Bratko Faculty of Computer and Information Science, University of Ljubljana, Slovenia Abstract. The main advantage of unordered classiflcation rules is in their power to spot and explain local regularities. However, using them in classiflcation often poses problems due to con°icts between rules, when
sortiert nach Relevanz / Datum