Network embedding plays an important role in various real-world applications. Most traditional algorithms focus on the topological structure while ignore the information from node attributes. The attributed information is potentially valuable to network embedding. To solve this problem, we propose a deep learning model named Extractive Convolutional Adversarial Network (ECAN) for network embedding. This model aims to extract the latent representations from the topological structure, the attributed information, and labels via three components. In the first part, ECAN extracts features from the topological structure and the attributed information of nodes separately. The second part is a prediction model, which aims to exploit labels of vertices. The third part is a convolutional adversarial model. We train it to distinguish the extractive features which are generated by the hidden layers in the extractive network from either the attributed information or the topological structure. Experiments on six real-world datasets demonstrate the effectiveness of ECAN when compared with state-of-the-art embedding algorithms.
|Number of pages||20|
|Journal||World Wide Web|
|Early online date||23 Nov 2019|
|Publication status||Published - May 2020|
Bibliographical noteThis work was supported by Top-Up Fund (TFG-04) and Seed Fund (SFG-10) for General Research Fund / Early Career Scheme and Interdisciplinary Research Scheme of the Dean’s Research Fund 2018-19 (FLASS/DRF/IDS-3), Departmental Collaborative Research Fund 2019 (MIT/DCRF-R2/18-19), Funding Support to General Research Fund Proposal (RG 39/2019-2020R) and the Internal Research Grant (RG 90/2018-2019R) of The Education University of Hong Kong, and LEO Dr David P. Chan Institute of Data Science, Lingnan University, Hong Kong. This work was also supported by the National Key R&D Program of China (2018YFB1004404), Key R&D Program of Guangdong Province (2018B010107005), and National Natural Science Foundation of China (U1711262, U1501252, U1611264, U1711261). This article is an extended journal version of a conference paper published at BESC 2018 . Some contents from the conference version are re-used in this journal article as this article is a follow-up work of the conference paper. The new contents of this article are more than 70% according to the regulation of the published journal. The new contents can be summarized in the following aspects: (1) To overcome the drawback that the training process of our basic method may be unstable, we propose a new method which exploits the idea of convolution in the network structure; (2) We add three metrics (i.e., Precision, Recall, and F1 score) to evaluate the effectiveness of different models comprehensively; (3) To test the influence of the kernel size in the hidden layers of the extractive model, the prediction model, and the convolutional adversarial model, we set this parameter to 1 and 3 for comparison; (4) Some recently developed approaches are included as baselines; (5) We conduct in-depth experiments on the proposed model to validate the effectiveness of the employed dimension reduction strategy and classifier.
- Attributed network
- Convolutional neural network
- Generative adversarial network
- Network embedding