Timestamp

Machine learning

Chemistry

Artificial intelligence

Linkage (software)

Embedding

Biochemistry

Physics

Dynamic network analysis

Gene

Optics

Quantum mechanics

Network formation

Data mining

Network simulation

Network architecture

Computer science

Distributed computing

World Wide Web

Theoretical computer science

Focus (optics)

Computer network

Network analysis

IEEE Access

Network embedding task aims at learning low-dimension latent representations of vertices while preserving the structure of a network simultaneously. Most existing network embedding methods mainly focus on static networks, which extract and condense the network information without temporal information. However, in the real world, networks keep evolving, where the linkage states between the same vertex pairs at consequential timestamps have very close correlations. In this paper, we propose to study the network embedding problem and focus on modeling the linkage evolution in the dynamic network setting. To address this problem, we propose a deep dynamic network embedding method. More specifically, the method utilizes the historical information obtained from the network snapshots at past timestamps to learn latent representations of the future network. In the proposed embedding method, the objective function is carefully designed to incorporate both the network internal and network dynamic transition structures. Extensive empirical experiments prove the effectiveness of the proposed model on various categories of real-world networks, including a human contact network, a bibliographic network, and e-mail networks. Furthermore, the experimental results also demonstrate the significant advantages of the method compared with both the state-of-the-art embedding techniques and several existing baseline methods.

Deep Dynamic Network Embedding for Link Prediction