A Linear Time Active Learning Algorithm for Link Classification -- Full Version --

TL;DR

This paper introduces a highly efficient active learning algorithm for link classification in signed networks, achieving near-optimal mistake bounds with low query complexity and fast running time.

Contribution

The paper proposes a novel linear-time active learning algorithm for link classification that is theoretically optimal within a constant factor under a stochastic model.

Findings

Achieves near-optimal mistake bounds with O(|V|^{3/2}) queries.

Runs in near-linear time relative to the size of the graph.

Provides a flexible algorithm with adjustable optimality factor.

Abstract

We present very efficient active learning algorithms for link classification in signed networks. Our algorithms are motivated by a stochastic model in which edge labels are obtained through perturbations of a initial sign assignment consistent with a two-clustering of the nodes. We provide a theoretical analysis within this model, showing that we can achieve an optimal (to whithin a constant factor) number of mistakes on any graph G = (V,E) such that |E| = \Omega(|V|^{3/2}) by querying O(|V|^{3/2}) edge labels. More generally, we show an algorithm that achieves optimality to within a factor of O(k) by querying at most order of |V| + (|V|/k)^{3/2} edge labels. The running time of this algorithm is at most of order |E| + |V|\log|V|.