Overview

Submodular optimization uses diminishing returns to model problems such as coverage, data selection, and allocating tasks or resources. Because exact optimization is often computationally hard, common methods include greedy selection, continuous relaxations with rounding, and local exchanges, with guarantees for constraints such as cardinality, knapsack, and matroids. Current work also studies non-monotone objectives and applications in language-model data selection, robotics, and benchmark design. A central tension is preserving strong approximation or regret guarantees while handling limited feedback, communication, changing constraints, and more general objectives.

Latest papers

61 papers7 papers in the last four weeks, down 22% on the four weeks before. 0.1% of all new papers.

May 24, 2023

Supermodular Rank: Set Function Decomposition and Optimization

We define the supermodular rank of a function on a lattice. This is the smallest number of terms needed to decompose it into a sum of supermodular functions. The supermodular summands are defined with respect to different partial orders. We characterize the maximum possible value of the supermodular rank and describe the functions with fixed supermodular rank. We analogously define the submodular rank. We use submodular decompositions to optimize set functions. Given a bound on the submodular rank of a set function, we formulate an algorithm that splits an optimization problem into submodular subproblems. We show that this method improves the approximation ratio guarantees of several algorithms for monotone set function maximization and ratio of set functions minimization, at a computation overhead that depends on the submodular rank.