What question did this study set out to answer?

To systematically categorize deep temporal point processes across modular components and establish a standardized empirical evaluation framework while advancing multi-type event relational discovery.

December 24, 2024IEEE Transactions on Knowledge and Data Engineering

An Extensive Survey With Empirical Studies on Deep Temporal Point Process

View Full Paper

Ask AI

Bookmark

Authors

HLHaitao LinFudan University CTCheng TanBeijing Academy of Artificial Intelligence LWLirong WuAnhui Jianzhu University

Discussion

Member takes

Implication

Comparative evaluation reveals improved fitting and prediction across asynchronous event sequences using modularized architectures, indicating the efficacy of latent Granger causality discovery.

Key Points

To systematically categorize deep temporal point processes across modular components and establish a standardized empirical evaluation framework while advancing multi-type event relational discovery.
Dismantled recent deep temporal point process models into four core modules—history encoding, conditional intensity function formulation, relational discovery, and optimization—and tested them under a unified learning strategy.
Extended history encoders and conditional intensity families while implementing a Granger causality discovery framework using discrete graph structure learning under variational inference.
Modular re-assembly enabled fair comparative benchmarking of existing history encoding and conditional intensity formulation methods under identical learning conditions.
The proposed latent graph discovery framework successfully identified multi-event relational structures while demonstrating superior event fitting and predictive accuracy over baseline models.

Cite This Study

Lin et al. (2024) studied this question.

synapsesocial.com/papers/6ab219c17fad5c2eb0c875a3 https://doi.org/10.1109/tkde.2024.3522114

View Full Paper

Ask AI

Bookmark

Implication

Comparative evaluation reveals improved fitting and prediction across asynchronous event sequences using modularized architectures, indicating the efficacy of latent Granger causality discovery.

An Extensive Survey With Empirical Studies on Deep Temporal Point Process

Discussion

Member takes

Implication

Key Points

Cite This Study

An Extensive Survey With Empirical Studies on Deep Temporal Point Process

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider