A Unifying Theory for the Reliability of Stochastic Programming Solutions Using Compromise Decisions

EasyChair Preprint 12439

Abstract

This paper studies the reliability of stochastic programming solutions using compromise decisions. A compromise decision is obtained by minimizing the aggregation of objective function approximations across the replications while regularizing the candidate decisions of all the replications, which we refer to as Compromise Decision problem. Rademacher average of families of functions is used to bound the sample complexity of the compromise decisions.

Keyphrases: Rademacher Average, sample average approximation, stochastic programming

@booklet{EasyChair:12439,
  author    = {Shuotao Diao and Suvrajeet Sen},
  title     = {A Unifying Theory for the Reliability of Stochastic Programming Solutions Using Compromise Decisions},
  howpublished = {EasyChair Preprint 12439},
  year      = {EasyChair, 2024}}