Multi-Scale attributed node embedding. (7th May 2021)

• Record Type: Journal Article
• Authors: Rozemberczki, Benedek; Allen, Carl; Sarkar, Rik
• Editors: Thilo Gross, xx
• Is Part Of: Journal of complex networks. Volume 9:Number 2(2021)
• Publication Date: 2021-05-07
• Subjects: node embedding -- node classification -- attributed network -- dimensionality reduction; Numerical analysis -- Periodicals; Computer networks -- Periodicals; Social networks -- Periodicals; 518.05
• DOI: 10.1093/comnet/cnab014
• Languages: English
• ISSNs: 2051-1310

Abstract

Abstract: We present network embedding algorithms that capture information about a node from the local distribution over node attributes around it, as observed over random walks following an approach similar to Skip-gram. Observations from neighbourhoods of different sizes are either pooled (AE) or encoded distinctly in a multi-scale approach (MUSAE). Capturing attribute-neighbourhood relationships over multiple scales is useful for a range of applications, including latent feature identification across disconnected networks with similar features. We prove theoretically that matrices of node-feature pointwise mutual information are implicitly factorized by the embeddings. Experiments show that our algorithms are computationally efficient and outperform comparable models on social networks and web graphs.