Loading [a11y]/accessibility-menu.js
Attributed Network Embedding with Data Distribution Adaptation | IEEE Conference Publication | IEEE Xplore

Attributed Network Embedding with Data Distribution Adaptation


Abstract:

Network embedding aims to learn the low-dimensional representations of nodes in networks and preserve the features of networks simultaneously. In this paper, we propose a...Show More

Abstract:

Network embedding aims to learn the low-dimensional representations of nodes in networks and preserve the features of networks simultaneously. In this paper, we propose a novel approach of network embedding, named Attributed Network Embedding with Data Distribution Adaptation (ANEDDA). In our model, we consider network structure and node attributes as two kinds of data which have different probability distributions. This is because that they come from different aspects or views and capture different kinds of features. However, because they come from the same network, there must exist some latent features which are related to both two kinds of information, and these features are significant to detect the community structure and analyze the network. We utilize a domain adaptation model named Transfer Component Analysis (TCA) to identify the latent common features. The proposed method is designed for unweighted networks (graphs), and it integrates the structure information and nodes attributes.
Date of Conference: 12-14 November 2018
Date Added to IEEE Xplore: 25 April 2019
ISBN Information:
Conference Location: Kaohsiung, Taiwan

References

References is not available for this document.