Tree Bounds for Sums of Bernoulli Random Variables: A Linear Optimization Approach

TL;DR

This paper develops a polynomial-time linear optimization method to compute tight probability bounds for sums of dependent Bernoulli variables structured as a tree, improving robustness over traditional univariate bounds.

Contribution

It introduces a novel linear optimization approach for tight probability bounds in tree-structured Bernoulli sums, extending to order statistics and more general variables.

Findings

Polynomial-time computation of bounds for tree-structured dependencies

Bounds are robust even when conditional independence assumptions are violated

Method outperforms univariate bounds in certain scenarios

Abstract

We study the problem of computing the tightest upper and lower bounds on the probability that the sum of $n$ dependent Bernoulli random variables exceeds an integer $k$. Under knowledge of all pairs of bivariate distributions denoted by a complete graph, the bounds are NP-hard to compute. When the bivariate distributions are specified on a tree graph, we show that tight bounds are computable in polynomial time using linear optimization. These bounds provide robust probability estimates when the assumption of conditional independence in a tree structured graphical model is violated. Generalization of the result to finding probability bounds of order statistic for more general random variables and instances where the bounds provide the most significant improvements over univariate bounds is also discussed in the paper.