Self-regularized causal structure discovery for trajectory-based networks

•Existing models rarely consider trajectories' time-varying properties.•cTVDBN reveals causal relationships among regions.•More reliable inferences can be made.•Approximate homotopy automates over-fitting control. Trajectory-based networks exhibit strong heterogeneous patterns amid human behavi...

Full description

Saved in:
Bibliographic Details
Published in:Journal of computer and system sciences 2016-06, Vol.82 (4), p.594-609
Main Authors: Chu, Victor W., Wong, Raymond K., Chen, Fang, Fong, Simon, Hung, Patrick C.K.
Format: Article
Language:English
Subjects:
Adaptive control systems
Approximation
Bayesian network
Causal structure discovery
Density-based clustering
Dynamics
Evolution
Human behavior
Mathematical models
Networks
Roads
Time-varying
Trajectories
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Existing models rarely consider trajectories' time-varying properties.•cTVDBN reveals causal relationships among regions.•More reliable inferences can be made.•Approximate homotopy automates over-fitting control. Trajectory-based networks exhibit strong heterogeneous patterns amid human behaviors. We propose a notion of causal time-varying dynamic Bayesian network (cTVDBN) to efficiently discover such patterns. While asymmetric kernels are used to make the model better adherence to causal principles, the variations of network connectivities are addressed by an adaptive over-fitting control. Compact regularization paths are obtained by approximate homotopy to make the solution tractable. In our experiments, cTVDBN structure discovery has successfully revealed the evolution of time-varying relationships in a ring road system, and provided insights for plausible road structure improvements from a traffic flow dataset.
ISSN:0022-0000
1090-2724
DOI:10.1016/j.jcss.2015.10.004