Graph-Based Transduction with Confidence

We present a new multi-class graph-based transduction algorithm. Examples are associated with vertices in an undirected weighted graph and edge weights correspond to a similarity measure between examples. Typical algorithms in such a setting perform label propagation between neighbours, ignoring the quality, or estimated quality, in the labeling of various nodes. We introduce an additional quantity of confidence in label assignments, and learn them jointly with the weights, while using them to dynamically tune the influence of each vertex on its neighbours. We cast learning as a convex optimization problem, and derive an efficient iterative algorithm for solving it. Empirical evaluations on seven NLP data sets demonstrate our algorithm improves over other state-of-the-art graph-based transduction algorithms.