Saved in:
Bibliographic Details
Main Authors: Nguyen, Khai, Ellinas, Petros, Bhagavathula, Anvita, Donti, Priya L.
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.05495
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914545593745408
author Nguyen, Khai
Ellinas, Petros
Bhagavathula, Anvita
Donti, Priya L.
author_facet Nguyen, Khai
Ellinas, Petros
Bhagavathula, Anvita
Donti, Priya L.
contents To scale optimization and simulation, prior work has explored training machine-learning surrogates that map problem parameters to solutions inexpensively at inference time. Unfortunately, commonly used approaches, including supervised and self-supervised learning with either soft or hard feasibility enforcement, face inherent challenges such as reliance on expensive high-quality labels or difficult optimization landscapes. To address their trade-offs, we propose a novel framework that collects "cheap" imperfect labels, performs supervised model pretraining with a merit loss-based termination scheme, and finally refines the model through self-supervised learning to improve final performance. Empirical validation across challenging domains -- including nonconvex constrained optimization, power-grid operation, and stiff dynamical systems -- shows that this three-stage strategy yields faster convergence; improved accuracy, feasibility, and optimality; and up to 59x reductions in total offline computational cost. We further analyze why and when our framework improves surrogate model training, finding that (i) merit loss is an informative signal and (ii) only small numbers of cheap, inexact labels are needed to place the model in a favorable regime for self-supervised learning.
format Preprint
id arxiv_https___arxiv_org_abs_2603_05495
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Cheap Thrills: Effective Amortized Optimization Using Inexpensive Labels
Nguyen, Khai
Ellinas, Petros
Bhagavathula, Anvita
Donti, Priya L.
Machine Learning
Optimization and Control
To scale optimization and simulation, prior work has explored training machine-learning surrogates that map problem parameters to solutions inexpensively at inference time. Unfortunately, commonly used approaches, including supervised and self-supervised learning with either soft or hard feasibility enforcement, face inherent challenges such as reliance on expensive high-quality labels or difficult optimization landscapes. To address their trade-offs, we propose a novel framework that collects "cheap" imperfect labels, performs supervised model pretraining with a merit loss-based termination scheme, and finally refines the model through self-supervised learning to improve final performance. Empirical validation across challenging domains -- including nonconvex constrained optimization, power-grid operation, and stiff dynamical systems -- shows that this three-stage strategy yields faster convergence; improved accuracy, feasibility, and optimality; and up to 59x reductions in total offline computational cost. We further analyze why and when our framework improves surrogate model training, finding that (i) merit loss is an informative signal and (ii) only small numbers of cheap, inexact labels are needed to place the model in a favorable regime for self-supervised learning.
title Cheap Thrills: Effective Amortized Optimization Using Inexpensive Labels
topic Machine Learning
Optimization and Control
url https://arxiv.org/abs/2603.05495