Unsupervised Learning of Hybrid Latent Dynamics: A Learn-to-Identify Framework

arXiv:2403.08194v1 Announce Type: new
Abstract: Modern applications increasingly require unsupervised learning of latent dynamics from high-dimensional time-series. This presents a significant challenge of identifiability: many abstract latent representations may reconstruct observations, yet do they guarantee an adequate identification of the governing dynamics? This paper investigates this challenge from two angles: the use of physics inductive bias specific to the data being modeled, and a learn-to-identify strategy that separates forecasting objectives from the data used for the identification. We combine these …

abstract applications arxiv challenge cs.lg dynamics framework hybrid identification identify learn modern modern applications paper series stat.ml type unsupervised unsupervised learning