Abstract

We organized a machine learning challenge on “active learning”, addressing problems where labeling data is expensive, but large amounts of unlabeled data are available at low cost. Examples include handwriting and speech recognition, document classification, vision tasks, drug design using recombinant molecules and protein engineering. The algorithms may place a limited number of queries to get new sample labels. The design of the challenge and its results are summarized in this paper and the best contributions made by the participants are included in these proceedings. The website of the challenge remains open as a resource for students and researchers (http://clopinet.com/al).