Efficient CoTraining of Linear Separators under Weak Dependence
[edit]
Proceedings of the 2017 Conference on Learning Theory, PMLR 65:302318, 2017.
Abstract
We develop the first polynomialtime algorithm for cotraining of homogeneous linear separators under \em weak dependence, a relaxation of the condition of independence given the label. Our algorithm learns from purely unlabeled data, except for a single labeled example to break symmetry of the two classes, and works for any data distribution having an inversepolynomial margin and with center of mass at the origin.
Related Material


