Open Peer Review. Open Publishing. Open Access. Open Discussion. Open Directory. Open Recommendations. Open API. Open Source.
Universum Prescription: Regularization using Unlabeled Data
Xiang Zhang, Yann LeCun
Feb 14, 2016 (modified: Feb 14, 2016)ICLR 2016 workshop submissionreaders: everyone
Abstract:This paper shows that simply prescribing "none of the above" labels to unlabeled data has a beneficial regularization effect to supervised learning. We call it universum prescription by the fact that the prescribed labels cannot be one of the supervised labels. In spite of its simplicity, universum prescription obtained competitive results in training deep convolutional networks for CIFAR-10, CIFAR-100 and STL-10 datasets. A qualitative justification of these approaches using Rademacher complexity is presented. The effect of a regularization parameter -- probability of sampling from unlabeled data -- is also studied empirically.
Conflicts:nyu.edu, cs.nyu.edu, tju.edu.cn
Enter your feedback below and we'll get back to you as soon as possible.