A Spatiotemporal Probabilistic Graphical Model Based on Adaptive Expectation-Maximization Attention for Individual Trajectory Reconstruction Considering Incomplete Observations

Xuan Sun ^1,2, Jianyuan Guo ¹, Yong Qin ²

⁰School of Traffic and Transportation, Beijing Jiaotong University, No. 3 Shangyuancun, Haidian District, Beijing 100044, China.

Entropy (basel, Switzerland) +

|

May 24, 2024

View abstract on PubMed

Summary

This study reconstructs urban rail transit trajectories using a novel spatiotemporal probabilistic graphical model with adaptive expectation maximization attention (STPGM-AEMA). The method accurately recovers missing trajectory data, improving operational strategies and personalized recommendations.

Area Of Science

Urban transportation systems
Data science and analytics
Probabilistic modeling

Background

Accurate individual trajectory data is crucial for urban rail transit operations.
Existing methods struggle to infer missing trajectory information from limited Automatic Fare Collection (AFC) and Automatic Vehicle Location (AVL) data.

Purpose Of The Study

To propose a novel method for reconstructing individual trajectories in urban rail transit.
To enhance the accuracy of inferring missing spatiotemporal information from incomplete trajectory data.

Main Methods

Developed a spatiotemporal probabilistic graphical model based on adaptive expectation maximization attention (STPGM-AEMA).
Incorporated data mining and combinatorial enumeration to identify potential train and egress time alternatives.
Utilized global and local potential variables for inference of unknown trajectory events.
Employed an attention mechanism-enhanced expectation-maximization algorithm for maximum likelihood estimation.

Main Results

The STPGM-AEMA method achieved over 95% accuracy in recovering missing trajectory information.
Demonstrated at least a 15% improvement in accuracy compared to traditional methods like PTAM-MLE and MPTAM-EM.
Validated effectiveness using origin-destination pair datasets and real individual trajectory tracking data.

Conclusions

The STPGM-AEMA method significantly enhances the reconstruction of individual urban rail transit trajectories.
This advancement offers improved capabilities for operational strategy adjustment, personalized recommendations, and emergency decision-making.

Keywords:

attention mechanism expectation-maximization algorithm probabilistic graphical model trajectory prediction urban rail transit