Outperformance Inverse Optimization: Learning Objective Functions that Outperform Agent Decisions
Abstract
Inverse optimization estimates the weights of an objective function that explain observed decisions as optimal solutions, and is used in a variety of fields. For mixed-integer linear programs (MILPs), existing methods aim to reproduce the observations as optimal solutions, and thus learn compromise weights when the observations are suboptimal. We propose outperformance inverse optimization, which instead seeks weights that induce, at each state, an optimal solution outperforming the observed action in every component. We give a loss function that can be evaluated with forward-problem oracles alone and is thus applicable to MILPs, together with gradient-based and DC optimization algorithms for minimizing it. For weights inducing a unique outperforming optimal solution at all observations, we prove that the probability of failing to induce such a solution at a new state (the generalization error) is bounded by a quantity inversely proportional to the number of observations, and that this bound is tight in the number of observations up to logarithmic factors. In experiments on synthetic and real data, the proposed methods improve the prediction of solutions outperforming the actions over existing methods.
Related
- A sequential quadratic programming method for optimization with expectation objective functions, deterministic inequality constraints and robust subproblems
- Synthesis of the Transient Behaviour of Optimization Algorithms for Strongly Convex Objective Functions
- Simple Extensions of Single-Objective Acquisition Functions and Hedge Strategies for Multi-Objective Bayesian Optimization
- Differential Properties of Objective and Value Functions for Discounted Markov Decision Processes with Finite State and Action Sets
- A connectedness result for convex polynomial multi-objective optimization without boundedness
BibTeX
@misc{kitaoka2026outperformance,
title = {{Outperformance Inverse Optimization: Learning Objective Functions that Outperform Agent Decisions}},
author = {Akira Kitaoka},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.09890},
eprint = {2610.09890},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.09890},
}
RIS
TY - UNPB
TI - Outperformance Inverse Optimization: Learning Objective Functions that Outperform Agent Decisions
AU - Akira Kitaoka
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2610.09890
UR - https://arxiv.org/abs/2610.09890
AB - Inverse optimization estimates the weights of an objective function that explain observed decisions as optimal solutions, and is used in a variety of fields. For mixed-integer linear programs (MILPs), existing methods aim to reproduce the observations as optimal solutions, and thus learn compromise weights when the observations are suboptimal. We propose outperformance inverse optimization, which instead seeks weights that induce, at each state, an optimal solution outperforming the observed action in every component. We give a loss function that can be evaluated with forward-problem oracles alone and is thus applicable to MILPs, together with gradient-based and DC optimization algorithms for minimizing it. For weights inducing a unique outperforming optimal solution at all observations, we prove that the probability of failing to induce such a solution at a new state (the generalization error) is bounded by a quantity inversely proportional to the number of observations, and that this bound is tight in the number of observations up to logarithmic factors. In experiments on synthetic and real data, the proposed methods improve the prediction of solutions outperforming the actions over existing methods.
ER -