Robust Adaptive Submodular Maximization

Summary: This paper studies the non-monotone adaptive submodular maximization problem and shows that the adaptive random greedy algorithm can achieve a 1/e approximation ratio under adaptive submodularity. Additionally, a linear-time algorithm is developed for non-monotone adaptive submodular maximization with a 1/e - epsilon approximation ratio. Furthermore, a faster algorithm is proposed for the monotone case with a 1 - 1/e - epsilon approximation ratio in expectation.