Great progresses have been made by researchers for the mutual devleopment between physically-based models (mechanism models) and Artificial intelligence (AI for short). Several typical milestones r...

“Inspired from the universal approximation theorem, neural networks with a single layer could not only approximates continuous functions, but also continuous functional or operators”1 Concepts fu...

SIMILAR to the former blog, the accurately learnt neural networks could also be beneficial when architectures are already known. Data-driven discovery for inverse problems Upon solving inverse pro...

Trained to solve supervised learning tasks, Physic-informed neural networks (PINNs) are defined for two classes of problems: data-driven solution and data-driven discovery of partial differential e...

Concepts in H2M Parameters in hybrid models are categorized into ML parameters and physical parameters1. Thus, hybrid modeling are grouped into parameteric (both learned) and non-parameteric (the ...