“Bring Your Own Greedy”+Max: Near-Optimal 1/2-Approximations for Submodular Knapsack
Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, PMLR 108:3263-3274, 2020.
The problem of selecting a small-size representative summary of a large dataset is a cornerstone of machine learning, optimization and data science. Motivated by applications to recommendation systems and other scenarios with query-limited access to vast amounts of data, we propose a new rigorous algorithmic framework for a standard formulation of this problem as a submodular maximization subject to a linear (knapsack) constraint. Our framework is based on augmenting all partial Greedy solutions with the best additional item. It can be instantiated with negligible overhead in any model of computation, which allows the classic greedy algorithm and its variants to be implemented. We give such instantiations in the offline Gready+Max, multi-pass streaming Sieve+Max and distributed Distributed Sieve+Max settings. Our algorithms give ($1/2-\eps$)-approximation with most other key parameters of interest being near-optimal. Our analysis is based on a new set of first-order linear differential inequalities and their robust approximate versions. Experiments on typical datasets (movie recommendations, influence maximization) confirm scalability and high quality of solutions obtained via our framework. Instance-specific approximations are typically in the 0.6-0.7 range and frequently beat even the $(1-1/e) \approx 0.63$ worst-case barrier for polynomial-time algorithms.