Structured Temporal Inference in State-Space Models

Hamidreza Hashempoor
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:2602-2610, 2026.

Abstract

We propose a framework for structured temporal inference in nonlinear state-space models (SSMs) with hybrid latent dynamics that mix discrete and continuous variables. Our method follows a two-stage inference: continuous states are estimated via Kalman inspired updates, while discrete variables are sampled by a neural model conditioned on these states, avoiding explicit Markov assumptions. To handle instabilities arising from recurrent dynamics, we introduce stabilization approach, and train all components jointly using surrogate gradient estimators that support REINFORCE-style updates. This design achieves SOTA results across synthetic and real-world datasets, in state estimation, regime detection, and imputation under noise and partial observability.

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-hashempoor26a, title = { Structured Temporal Inference in State-Space Models }, author = {Hashempoor, Hamidreza}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {2602--2610}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/hashempoor26a/hashempoor26a.pdf}, url = {https://proceedings.mlr.press/v300/hashempoor26a.html}, abstract = { We propose a framework for structured temporal inference in nonlinear state-space models (SSMs) with hybrid latent dynamics that mix discrete and continuous variables. Our method follows a two-stage inference: continuous states are estimated via Kalman inspired updates, while discrete variables are sampled by a neural model conditioned on these states, avoiding explicit Markov assumptions. To handle instabilities arising from recurrent dynamics, we introduce stabilization approach, and train all components jointly using surrogate gradient estimators that support REINFORCE-style updates. This design achieves SOTA results across synthetic and real-world datasets, in state estimation, regime detection, and imputation under noise and partial observability. } }
Endnote
%0 Conference Paper %T Structured Temporal Inference in State-Space Models %A Hamidreza Hashempoor %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-hashempoor26a %I PMLR %P 2602--2610 %U https://proceedings.mlr.press/v300/hashempoor26a.html %V 300 %X We propose a framework for structured temporal inference in nonlinear state-space models (SSMs) with hybrid latent dynamics that mix discrete and continuous variables. Our method follows a two-stage inference: continuous states are estimated via Kalman inspired updates, while discrete variables are sampled by a neural model conditioned on these states, avoiding explicit Markov assumptions. To handle instabilities arising from recurrent dynamics, we introduce stabilization approach, and train all components jointly using surrogate gradient estimators that support REINFORCE-style updates. This design achieves SOTA results across synthetic and real-world datasets, in state estimation, regime detection, and imputation under noise and partial observability.
APA
Hashempoor, H.. (2026). Structured Temporal Inference in State-Space Models . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:2602-2610 Available from https://proceedings.mlr.press/v300/hashempoor26a.html.

Related Material