Multi-task Modeling for Engineering Applications with Sparse Data

TL;DR

This paper presents a Multi-Task Gaussian Processes framework designed for engineering applications with multi-source, multi-fidelity data, effectively handling data scarcity and task correlations to improve predictions and reduce costs.

Contribution

The paper introduces a novel MTGP framework that models inter-task relationships across multiple fidelity levels, enhancing predictive accuracy in data-sparse engineering contexts.

Findings

Improved predictive performance across diverse engineering scenarios.

Reduced computational costs compared to traditional methods.

Validated robustness and scalability of the framework.

Abstract

Modern engineering and scientific workflows often require simultaneous predictions across related tasks and fidelity levels, where high-fidelity data is scarce and expensive, while low-fidelity data is more abundant. This paper introduces an Multi-Task Gaussian Processes (MTGP) framework tailored for engineering systems characterized by multi-source, multi-fidelity data, addressing challenges of data sparsity and varying task correlations. The proposed framework leverages inter-task relationships across outputs and fidelity levels to improve predictive performance and reduce computational costs. The framework is validated across three representative scenarios: Forrester function benchmark, 3D ellipsoidal void modeling, and friction-stir welding. By quantifying and leveraging inter-task relationships, the proposed MTGP framework offers a robust and scalable solution for predictive modeling in domains with significant computational and experimental costs, supporting informed decision-making and efficient resource utilization.