[2603.22886] Conditionally Identifiable Latent Representation for Multivariate Time Series with Structural Dynamics

Computer Science > Machine Learning arXiv:2603.22886 (cs) [Submitted on 24 Mar 2026] Title:Conditionally Identifiable Latent Representation for Multivariate Time Series with Structural Dynamics Authors:Minkey Chang, Jae-Young Kim View a PDF of the paper titled Conditionally Identifiable Latent Representation for Multivariate Time Series with Structural Dynamics, by Minkey Chang and 1 other authors View PDF HTML (experimental) Abstract:We propose the Identifiable Variational Dynamic Factor Model (iVDFM), which learns latent factors from multivariate time series with identifiability guarantees. By applying iVAE-style conditioning to the innovation process driving the dynamics rather than to the latent states, we show that factors are identifiable up to permutation and component-wise affine (or monotone invertible) transformations. Linear diagonal dynamics preserve this identifiability and admit scalable computation via companion-matrix and Krylov methods. We demonstrate improved factor recovery on synthetic data, stable intervention accuracy on synthetic SCMs, and competitive probabilistic forecasting on real-world benchmarks. Comments: Subjects: Machine Learning (cs.LG); General Finance (q-fin.GN); Statistical Finance (q-fin.ST) Cite as: arXiv:2603.22886 [cs.LG]   (or arXiv:2603.22886v1 [cs.LG] for this version)   https://doi.org/10.48550/arXiv.2603.22886 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Minkey Chang [view ema...