vix.ing · top · new · best · stats · spec

A Reduction for Optimizing Lattice Submodular Functions with Diminishing\n Returns

2016/06/27 by Alina Ene, Ene, Alina, Huy L. Nguyên +1 · 4 citations
Computer Science · #Advanced Algebra and Logic #Advanced Graph Theory Research #Artificial Intelligence (cs.AI) #Complexity and Algorithms in Graphs #Cryptography and Data Security #Data Structures and Algorithms (cs.DS) #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · pdf · doi:10.48550/arxiv.1606.08362

openalex publication_date 2016/06/27 · openalex created_date 2022/09/04 · openalex updated_date 2026/07/28

Abstract

A function f: \ℤ+E \→ \ℝ+ is DR-submodular if\nit satisfies f( bf x + \χi) -f ( bf x) \≥ f( bf y + \χi) - f( bf\ny) for all bf x\≤ bf y, i\∈ E. Recently, the problem of maximizing\na DR-submodular function f: \ℤ+E \→ \ℝ+ subject\nto a budget constraint \‖ bf x\‖1 \≤ B as well as additional constraints\nhas received significant attention citeSKIK14,SY15,MYK15,SY16.\n In this note, we give a generic reduction from the DR-submodular setting to\nthe submodular setting. The running time of the reduction and the size of the\nresulting submodular instance depends only \logarithmically on B. Using\nthis reduction, one can translate the results for unconstrained and constrained\nsubmodular maximization to the DR-submodular setting for many types of\nconstraints in a unified manner.\n

Citations

Cited by

Related