Machine Learning Approaches for Interactive Verification

Interactive verification is a new problem, which is closely related to active learning, but aims to query as many positive instances as possible within some limited query budget. We point out the similarity between interactive verification and another machine learning problem called contextual bandit. The similarity allows us to design interactive verification approaches from existing contextual bandit approaches. We compare the performance of those approaches on interactive verification. In particular, we propose to adopt the upper confidence bound (UCB) algorithm, which has been widely used for the contextual bandit, to solve the interactive verification problem. Experiment results demonstrate that UCB reaches superior performance for interactive verification on many real-world datasets.