Adaptive submodular optimization, where a sequence of items is selected adaptively to optimize a submodular function, has been found to have many applications from sensor placement to active learning. In the current paper, we extend this work to the setting of multiple queries at each time step, where the set of available queries is randomly constrained. A primary contribution of this paper is to prove the first near optimal approximation bound for a greedy policy in this setting. A natural application of this framework is to crowd-sourced active learning problem where the set of available experts and examples might vary randomly. We instantiate the new framework for multi-label learning and evaluate it in multiple benchmark domains with promising results.

Computational complexity is still a challenging problem for intelligent systems operating in compound environments. To tackle it, an agent has to deal with perceptual information intelligently. In this paper, we propose an efficient and adaptive reasoning system based on Adaptive Logic Interpreter reasoning system, a mechanism for guiding inference through relational reinforcement learning, and a variation of Truth Maintenance Systems to speed up the inference. Relational reinforcement learning guides the inference toward the most rewarding parts of the knowledge base and truth maintenance system maintains beliefs, avoids repetitive inferences and reduces the state space. Empirical results demonstrate higher performance than the basic approach in terms of number of inferred instances, average reward, and average reward accuracy.