Distribution-Independent PAC Learning of Halfspaces with Massart Noise

Neural Information Processing Systems 

We study the problem of {\em distribution-independent} PAC learning of halfspaces in the presence of Massart noise.