We present and analyze an agnostic active learning algorithm that works without keeping a version space. This is unlike all previous approaches where a restricted set of candidate...
Alina Beygelzimer, Daniel Hsu, John Langford, Tong...
We give the first representation-independent hardness results for PAC learning intersections of halfspaces, a central concept class in computational learning theory. Our hardness ...
In this paper we briefly discuss the knowledge-on-demand (KOD) paradigm as it emerges from the current needs of the knowledge-based society. Basic requirements for on-demand learn...
Demetrios G. Sampson, Charalampos Karagiannidis, A...
: To make a new problem from the original one and to compare their solutions are promising activities to promote a learner to be aware of the structure of these problems. Especiall...
We address the problem of efficiently learning Naive Bayes classifiers under classconditional classification noise (CCCN). Naive Bayes classifiers rely on the hypothesis that the ...