Fast non-monotone submodular maximisation subject to a matroid constraint

Date published

2017-03-21

Free to read from

Supervisor/s

Journal Title

Journal ISSN

Volume Title

Publisher

Department

Type

Report

ISSN

Format

Citation

Pau Segui-Gasco and Hyo-Sang Shin. Fast non-monotone submodular maximisation subject to a matroid constraint

Abstract

In this work we present the first practical . 1 e −ǫ . -approximation algorithm to maximise a general non-negative submodular function subject to a matroid constraint. Our algorithm is based on combining the decreasing-threshold procedure of Badanidiyuru and Vondrak (SODA 2014) with a smoother version of the measured continuous greedy algorithm of Feldman et al. (FOCS 2011). This enables us to obtain an algorithm that requires O( nr2 ǫ4 . ¯ d+ ¯ d ¯ d .2 log2 ( n ǫ )) value oracle calls, where n is the cardinality of the ground set, r is the matroid rank, and ¯ d, ¯ d ∈R+ are the absolute values of the minimum and maximum marginal values that the function f can take i.e.: − ¯ d ≤fS(i) ≤ ¯ d, for all i ∈E and S ⊆E, where E is the ground set. The additional value oracle calls with respect to the work of Badanidiyuru and Vondrak come from the greater spread in the sampling of the multilinear extension that the possibility of negative marginal values introduce.

Description

Software Description

Software Language

Github

Keywords

DOI

Rights

This is the Author Accepted Manuscript. This article may be used for personal use only. The final published version (version of record) is available online at https://arxiv.org/abs/1703.06053

Relationships

Relationships

Supplements

Funder/s