Id: arxiv/2603.23299 Url: https://arxiv.org/abs/2603.23299 Tags: math.OC Score: 0 Title: Pruning for efficient deterministic global optimization over trained ReLU neural networks Domain: math.OC Source: arxiv Authors: Giacomo Lastrucci, Tanuj Karia, Victor Schulte, Dominik Bongartz, Artur M. Schweidtmann Reviews: 0 Subject: math.OC Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-03-24 Page: https://delphi.science/r/arxiv/2603.23299 Neural networks are increasingly used as surrogates in optimization problems to replace computationally expensive models. However, embedding ReLU neural networks in mathematical programs introduces significant computational challenges, particularly for deep and wide networks, due to both the formulation of the ReLU disjunction and the resulting large-scale optimization problem. This work investigates how pruning techniques can accelerate the solution of optimization problems with embedded neural networks, focusing on the mechanisms underlying the computational gains. We provide theoretical…