SurroFlow: A Flow-Based Surrogate Model for Parameter Space Exploration and Uncertainty Quantification

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Shen, Jingyi, Duan, Yuhan, Shen, Han-Wei
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917725453942784
author Shen, Jingyi
Duan, Yuhan
Shen, Han-Wei
author_facet Shen, Jingyi
Duan, Yuhan
Shen, Han-Wei
contents Existing deep learning-based surrogate models facilitate efficient data generation, but fall short in uncertainty quantification, efficient parameter space exploration, and reverse prediction. In our work, we introduce SurroFlow, a novel normalizing flow-based surrogate model, to learn the invertible transformation between simulation parameters and simulation outputs. The model not only allows accurate predictions of simulation outcomes for a given simulation parameter but also supports uncertainty quantification in the data generation process. Additionally, it enables efficient simulation parameter recommendation and exploration. We integrate SurroFlow and a genetic algorithm as the backend of a visual interface to support effective user-guided ensemble simulation exploration and visualization. Our framework significantly reduces the computational costs while enhancing the reliability and exploration capabilities of scientific surrogate models.
format Preprint
id arxiv_https___arxiv_org_abs_2407_12884
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SurroFlow: A Flow-Based Surrogate Model for Parameter Space Exploration and Uncertainty Quantification
Shen, Jingyi
Duan, Yuhan
Shen, Han-Wei
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Graphics
Human-Computer Interaction
Existing deep learning-based surrogate models facilitate efficient data generation, but fall short in uncertainty quantification, efficient parameter space exploration, and reverse prediction. In our work, we introduce SurroFlow, a novel normalizing flow-based surrogate model, to learn the invertible transformation between simulation parameters and simulation outputs. The model not only allows accurate predictions of simulation outcomes for a given simulation parameter but also supports uncertainty quantification in the data generation process. Additionally, it enables efficient simulation parameter recommendation and exploration. We integrate SurroFlow and a genetic algorithm as the backend of a visual interface to support effective user-guided ensemble simulation exploration and visualization. Our framework significantly reduces the computational costs while enhancing the reliability and exploration capabilities of scientific surrogate models.
title SurroFlow: A Flow-Based Surrogate Model for Parameter Space Exploration and Uncertainty Quantification
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Graphics
Human-Computer Interaction
url https://arxiv.org/abs/2407.12884