Optimath-X: An Explainable AI- Driven Mathematical Optimization Framework for Intelligent Decision -Making and Complex Problem Solving
Optimization algorithms in mathematics are key tools in the process of decision-making in engineering, logistics, and resources management, but traditional solvers operate like black boxes and provide no insights into how their solutions are obtained and on what grounds particular decisions are made. This lack of transparency is problematic for adopting the tool in contexts, where decision-makers have to be convinced before they make a move based on solver's suggestions. In this paper, we introduce an explainable AI-based optimization framework OptiMath-X, which unites surrogate-assisted metaheuristic search with explanation generation for mathematical optimization. The approach combines an evolutionary–local search solver with a consistent local explanation generator, which is constrained by problem structure and dependencies between variables. The experiments, conducted using benchmark problems and real-world instance from logistics domain show that our approach provides quality of solutions comparable to traditional solvers, and generated explanations are found useful by domain experts.
Authors
- N.Indrajith
- Vidhya K
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24412820
- Primary Topic
- Explainable Artificial Intelligence (XAI)
- Type
- article
- Field-Weighted Citation Impact
- 0.00