Interpretable System Identification
Black-box surrogates are fast but hard to trust. This line of work replaces them with identified models whose structure can be read and checked: sparse latent dynamics, port-Hamiltonian structure that guarantees passivity, and probabilistic coefficients that quantify uncertainty.
Key results are VENI, VINDy, VICI, a generative reduced-order modeling framework combining variational autoencoders with a probabilistic extension of SINDy, and the data-driven identification of latent port-Hamiltonian systems.
Jonas Kneifl
Postdoctoral Researcher
My research combines model order reduction, surrogate modeling, and machine learning — and asks what physics generative models learn.
