Label Propagation Through Linear Neighborhoods

Abstract

A novel semi-supervised learning approach is proposed based on a linear neighborhood model, which assumes that each data point can be linearly reconstructed from its neighborhood. Our algorithm, named <i>Linear Neighborhood Propagation (LNP)</i>, can propagate the labels from the labeled points to the whole dataset using these linear neighborhoods with sufficient smoothness. We also derive an easy way to extend <i>LNP</i> to out-of-sample data. Promising experimental results are presented for synthetic data, digit and text classification tasks.