Correlated random features for fast semi-supervised learning

This paper presents Correlated Nyström Views (XNV), a fast semi-supervised algorithm for regression and classification. The algorithm draws on two main ideas. First, it generates two views consisting of computationally inexpensive random features.