Department of Computer Science
Detecting multiple stochastic network motifs in network data
Network motif detection methods are known to be important for studying the structural properties embedded in network data. Extending them to stochastic ones help capture the interaction uncertainties in stochastic networks. In this paper, we propose a finite mixture model to detect multiple stochastic motifs in network data with the conjecture that interactions to be modeled in the motifs are of stochastic nature. Component-wise Expectation Maximization algorithm is employed so that both the optimal number of motifs and the parameters of their corresponding probabilistic models can be estimated. For evaluating the effectiveness of the algorithm, we applied the stochastic motif detection algorithm to both synthetic and benchmark datasets. Also, we discuss how the obtained stochastic motifs could help the domain experts to gain better insights on the over-represented patterns in the network data. © 2012 Springer-Verlag.
expectation maximization algorithm, finite mixture models, social networks, Stochastic motifs
Source Publication Title
Advances in knowledge discovery and data mining: 16th Pacific-Asia Conference, PAKDD 2012, Kuala Lumpur, Malaysia, May 29 – June 1, 2012, Proceedings, Part II
Tan, Pang-Ning ; Chawla, Sanjay ; Ho, Chin Kuan ; Bailey, James
Lecture Notes in Computer Science
Springer Berlin Heidelberg
Liu, Kai, William K. Cheung, and Jiming Liu. "Detecting multiple stochastic network motifs in network data." Advances in knowledge discovery and data mining: 16th Pacific-Asia Conference, PAKDD 2012, Kuala Lumpur, Malaysia, May 29 – June 1, 2012, Proceedings, Part II (2012): 205-217.