Multi-task Modeling for Engineering Applications with Sparse Data
This addresses predictive modeling challenges in engineering domains with high computational and experimental costs, but it is incremental as it builds on existing multi-task Gaussian process methods.
The paper tackles the problem of making simultaneous predictions across related tasks and fidelity levels in engineering applications where high-fidelity data is sparse and expensive, by introducing a Multi-Task Gaussian Processes framework that leverages inter-task relationships to improve predictive performance and reduce computational costs, validated on benchmarks like the Forrester function and 3D ellipsoidal void modeling.
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.