TU Darmstadt / ULB / TUbiblio

Dependent Relational Gamma Process Models for Longitudinal Networks

Yang, S. and Koeppl, H. :
Dependent Relational Gamma Process Models for Longitudinal Networks.
[Online-Edition: https://icml.cc/Conferences/2018/Schedule?showEvent=1942]
In: Thirty-fifth International Conference on Machine Learning, July 10-15, 2018, Stockholm, Denmark. In: Proceedings of Machine Learning Research , 80
[Conference or Workshop Item] , (2018)

Official URL: https://icml.cc/Conferences/2018/Schedule?showEvent=1942

Abstract

A probabilistic framework based on the covariate-dependent relational gamma process is developed to analyze relational data arising from longitudinal networks. The proposed framework characterizes networked nodes by nonnegative node-group memberships, which allow each node to belong to multiple latent groups simultaneously, and encodes edge probabilities between each pair of nodes using a Bernoulli Poisson link to the embedded latent space. Within the latent space, our framework models the birth and death dynamics of individual groups via a thinning function. Our framework also captures the evolution of individual node-group memberships over time using gamma Markov processes. Exploiting the recent advances in data augmentation and marginalization techniques, a simple and efficient Gibbs sampler is proposed for posterior computation. Experimental results on a simulation study and three real-world temporal network data sets demonstrate the model’s capability, competitive performance and scalability compared to state-of-the-art methods.

Item Type: Conference or Workshop Item
Erschienen: 2018
Creators: Yang, S. and Koeppl, H.
Title: Dependent Relational Gamma Process Models for Longitudinal Networks
Language: English
Abstract:

A probabilistic framework based on the covariate-dependent relational gamma process is developed to analyze relational data arising from longitudinal networks. The proposed framework characterizes networked nodes by nonnegative node-group memberships, which allow each node to belong to multiple latent groups simultaneously, and encodes edge probabilities between each pair of nodes using a Bernoulli Poisson link to the embedded latent space. Within the latent space, our framework models the birth and death dynamics of individual groups via a thinning function. Our framework also captures the evolution of individual node-group memberships over time using gamma Markov processes. Exploiting the recent advances in data augmentation and marginalization techniques, a simple and efficient Gibbs sampler is proposed for posterior computation. Experimental results on a simulation study and three real-world temporal network data sets demonstrate the model’s capability, competitive performance and scalability compared to state-of-the-art methods.

Journal or Publication Title: Proceedings of Machine Learning Research
Title of Book: Proceedings of Machine Learning Research (PMLR)
Volume: 80
Divisions: 18 Department of Electrical Engineering and Information Technology
18 Department of Electrical Engineering and Information Technology > Institute for Telecommunications > Bioinspired Communication Systems
18 Department of Electrical Engineering and Information Technology > Institute for Telecommunications
Event Title: Thirty-fifth International Conference on Machine Learning
Event Location: Stockholm, Denmark
Event Dates: July 10-15, 2018
Date Deposited: 01 Aug 2018 08:07
Official URL: https://icml.cc/Conferences/2018/Schedule?showEvent=1942
Export:

Optionen (nur für Redakteure)

View Item View Item