On Additive Approximate Submodularity

TL;DR

This paper investigates the closeness of approximately submodular functions to truly submodular functions, providing bounds and algorithms to adapt optimization techniques in machine learning contexts.

Contribution

It proves that approximately submodular functions are O(n^2) close to true submodular functions and offers an algorithmic approach for optimization adaptation.

Findings

Approximately submodular functions are O(n^2) close to submodular functions.

There is an ( ext{sqrt}(n)) lower bound on the distance to submodularity.

Contrast with approximate modularity and convexity cases.

Abstract

A real-valued set function is (additively) approximately submodular if it satisfies the submodularity conditions with an additive error. Approximate submodularity arises in many settings, especially in machine learning, where the function evaluation might not be exact. In this paper we study how close such approximately submodular functions are to truly submodular functions. We show that an approximately submodular function defined on a ground set of $n$ elements is $O(n^2)$ pointwise-close to a submodular function. This result also provides an algorithmic tool that can be used to adapt existing submodular optimization algorithms to approximately submodular functions. To complement, we show an $\Omega(\sqrt{n})$ lower bound on the distance to submodularity. These results stand in contrast to the case of approximate modularity, where the distance to modularity is a constant, and approximate convexity, where the distance to convexity is logarithmic.