Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces Dynamic Constraint Learning (DCL), a novel framework for constrained optimization that operates effectively without prior knowledge of constraint functions. By leveraging local surrogates that adapt to the data distribution, DCL solves optimization subproblems within a data-supported trust region, achieving comparable solution quality to global models while simplifying the optimization process. Experimental results on synthetic and real-world case studies validate the efficiency and effectiveness of DCL in constrained optimization scenarios.
Local surrogates in Dynamic Constraint Learning can match the performance of global models while simplifying the optimization landscape.
We propose Dynamic Constraint Learning (DCL), a data-driven framework for constrained optimization when constraint functions are unknown and cannot be queried during optimization. At each iteration, the method learns a local surrogate from nearby data and solves a subproblem within a data-supported trust region. Compared with offline global constraint learning, the approach uses local surrogates that adapt to the data distribution during optimization and can achieve solution quality comparable to that of global models while using simpler local models and smaller optimization subproblems. We demonstrate the performance of DCL on a synthetic test problem and two case studies from the literature.