A Comparative Review of Probabilistic Inference for Timescale Graphical Event Models

El Mokhtar Hribach, Julien Blanchard, Philippe Leray
Proceedings of the 13th International Conference on Probabilistic Graphical Models, PMLR 343:379-391, 2026.

Abstract

Timescale Graphical Event Models (TGEMs) provide a continuous-time probabilistic framework for representing dependencies among multivariate point processes using a directed graph, where each node corresponds to an event label and each edge specifies how occurrences of a parent event label influence the conditional intensity of a child event label over future time intervals, called timescales. Although TGEMs have been mainly investigated from a modeling and learning perspective, the practical behavior of sampling-based inference methods in these models remains insufficiently characterized. In this paper, we empirically compare three sampling strategies for probabilistic inference in TGEMs: forward sampling, Poisson Superposition Importance Sampling (PSIS), and auxiliary Gibbs sampling. We evaluate these methods on episodic sequence queries with incomplete and imprecise evidence. We examine the consistency of probability estimates, convergence behavior, and sensitivity to the rarity of the evidence. Our results provide a comparative experimental framework for assessing the strengths and limitations of each strategy and offer practical guidance for method selection in TGEM-based inference.

Cite this Paper

BibTeX
@InProceedings{pmlr-v343-hribach26a, title = {A Comparative Review of Probabilistic Inference for Timescale Graphical Event Models}, author = {Hribach, El Mokhtar and Blanchard, Julien and Leray, Philippe}, booktitle = {Proceedings of the 13th International Conference on Probabilistic Graphical Models}, pages = {379--391}, year = {2026}, editor = {Camps-Valls, Gustau and Leonelli, Manuele and Varando, Gherardo}, volume = {343}, series = {Proceedings of Machine Learning Research}, month = {09--11 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v343/main/assets/hribach26a/hribach26a.pdf}, url = {https://proceedings.mlr.press/v343/hribach26a.html}, abstract = {Timescale Graphical Event Models (TGEMs) provide a continuous-time probabilistic framework for representing dependencies among multivariate point processes using a directed graph, where each node corresponds to an event label and each edge specifies how occurrences of a parent event label influence the conditional intensity of a child event label over future time intervals, called timescales. Although TGEMs have been mainly investigated from a modeling and learning perspective, the practical behavior of sampling-based inference methods in these models remains insufficiently characterized. In this paper, we empirically compare three sampling strategies for probabilistic inference in TGEMs: forward sampling, Poisson Superposition Importance Sampling (PSIS), and auxiliary Gibbs sampling. We evaluate these methods on episodic sequence queries with incomplete and imprecise evidence. We examine the consistency of probability estimates, convergence behavior, and sensitivity to the rarity of the evidence. Our results provide a comparative experimental framework for assessing the strengths and limitations of each strategy and offer practical guidance for method selection in TGEM-based inference.} }
Endnote
%0 Conference Paper %T A Comparative Review of Probabilistic Inference for Timescale Graphical Event Models %A El Mokhtar Hribach %A Julien Blanchard %A Philippe Leray %B Proceedings of the 13th International Conference on Probabilistic Graphical Models %C Proceedings of Machine Learning Research %D 2026 %E Gustau Camps-Valls %E Manuele Leonelli %E Gherardo Varando %F pmlr-v343-hribach26a %I PMLR %P 379--391 %U https://proceedings.mlr.press/v343/hribach26a.html %V 343 %X Timescale Graphical Event Models (TGEMs) provide a continuous-time probabilistic framework for representing dependencies among multivariate point processes using a directed graph, where each node corresponds to an event label and each edge specifies how occurrences of a parent event label influence the conditional intensity of a child event label over future time intervals, called timescales. Although TGEMs have been mainly investigated from a modeling and learning perspective, the practical behavior of sampling-based inference methods in these models remains insufficiently characterized. In this paper, we empirically compare three sampling strategies for probabilistic inference in TGEMs: forward sampling, Poisson Superposition Importance Sampling (PSIS), and auxiliary Gibbs sampling. We evaluate these methods on episodic sequence queries with incomplete and imprecise evidence. We examine the consistency of probability estimates, convergence behavior, and sensitivity to the rarity of the evidence. Our results provide a comparative experimental framework for assessing the strengths and limitations of each strategy and offer practical guidance for method selection in TGEM-based inference.
APA
Hribach, E.M., Blanchard, J. & Leray, P.. (2026). A Comparative Review of Probabilistic Inference for Timescale Graphical Event Models. Proceedings of the 13th International Conference on Probabilistic Graphical Models, in Proceedings of Machine Learning Research 343:379-391 Available from https://proceedings.mlr.press/v343/hribach26a.html.

Related Material