Manifold Boost: Stagewise Function Approximation for Fully-, Semi- and Un-supervised Learning

We describe a manifold learning framework that naturally accommodates supervised learning manifold learning, partially supervised learning and unsupervised clustering as particular cases. Our method c