Multi-Fidelity Residual Neural Processes for Scalable Surrogate Modeling

arXiv:2402.18846v1 Announce Type: new
Abstract: Multi-fidelity surrogate modeling aims to learn an accurate surrogate at the highest fidelity level by combining data from multiple sources. Traditional methods relying on Gaussian processes can hardly scale to high-dimensional data. Deep learning approaches utilize neural network based encoders and decoders to improve scalability. These approaches share encoded representations across fidelities without including corresponding decoder parameters. At the highest fidelity, the representations are decoded with different parameters, making the shared information inherently inaccurate. This …

abstract arxiv cs.lg data deep learning fidelity gaussian processes learn modeling multiple network neural network processes residual scalability scalable scale type