What question did this study set out to answer?

To review recent advances in deep spatiotemporal point processes, focusing on deep influence kernel frameworks that merge neural network expressiveness with interpretable event propagation models.

December 22, 2025Annual Review of Statistics and Its Application

Deep Spatiotemporal Point Processes: Advances and New Directions

View Full Paper

Ask AI

Bookmark

Authors

XCXiuyuan ChengDuke University ZDZheng DongNorth Sichuan Medical University YXYao XieGeorgia Institute of Technology

Discussion

Member takes

Implication

Methodological review demonstrates deep influence kernel modeling for spatiotemporal point processes, highlighting improved balance between neural expressive power and statistical interpretability.

Key Points

To review recent advances in deep spatiotemporal point processes, focusing on deep influence kernel frameworks that merge neural network expressiveness with interpretable event propagation models.
Reviewed deep neural architectures integrated into point processes, specifically contrasting direct conditional intensity modeling with deep influence kernel methods.
Analyzed structural modeling elements including functional basis decomposition, graph neural networks for spatial topologies, and likelihood-based versus likelihood-free estimation.
Evaluated computational scalability approaches for large-scale event data and theoretical frameworks concerning kernel identifiability.
Deep influence kernels successfully preserve statistical interpretability through explicit transmission kernels while leveraging neural networks to model nonstationary, heterogeneous dynamics.
Practical deployments in seismology (earthquake aftershocks), criminology, and epidemiology (clinical sepsis onset) validate the framework's versatility and predictive capacity.
Identified critical open challenges and future research paths centered on improving model explainability, identifiability guarantees, and computational efficiency.

Cite This Study

Cheng et al. (2025) studied this question.

synapsesocial.com/papers/6a157967814bf8ec9a4ea44a https://doi.org/10.1146/annurev-statistics-042324-040052

View Full Paper

Ask AI

Bookmark

Implication

Methodological review demonstrates deep influence kernel modeling for spatiotemporal point processes, highlighting improved balance between neural expressive power and statistical interpretability.

Deep Spatiotemporal Point Processes: Advances and New Directions

Discussion

Member takes

Implication

Key Points

Cite This Study

Deep Spatiotemporal Point Processes: Advances and New Directions

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider