Adaptive Sequence Submodularity

Abstract

In many machine learning applications, one needs to interactively select a sequence of items (e.g., recommending movies based on a user's feedback) or make sequential decisions in a certain order (e.g., guiding an agent through a series of states). Not only do sequences already pose a dauntingly large search space, but we must also take into account past observations, as well as the uncertainty of future outcomes. Without further structure, finding an optimal sequence is notoriously challenging, if not completely intractable. In this paper, we view the problem of adaptive and sequential decision making through the lens of submodularity and propose an adaptive greedy policy with strong theoretical guarantees. Additionally, to demonstrate the practical utility of our results, we run experiments on Amazon product recommendation and Wikipedia link prediction tasks.

References in corpus (6)

• Sequence to Sequence Learning with Neural Networks
• Near-optimal Nonmyopic Value of Information in Graphical Models
• Beyond $1/2$-Approximation for Submodular Maximization on Massive Data Streams
• Robust Submodular Maximization: A Non-Uniform Partitioning Approach
• Submodular Functions: from Discrete to Continous Domains
• Beyond Adaptive Submodularity: Approximation Guarantees of Greedy Policy with Adaptive Submodularity Ratio

Cited by in corpus (1)

• Submodularity in Action: From Machine Learning to Signal Processing Applications