@article {Wang247577, author = {Xu-Wen Wang and Yize Chen and Yang-Yu Liu}, title = {Link Prediction through Deep Learning}, elocation-id = {247577}, year = {2018}, doi = {10.1101/247577}, publisher = {Cold Spring Harbor Laboratory}, abstract = {Inferring missing links or predicting future ones based on the currently observed network is known as the link prediction problem, which has tremendous real-world applications. Indeed, a successful link prediction method will substantially reduce the experimental effort required to establish the topology of a network (such as the protein-protein interaction network or the drug-target interaction network). It will also accelerate mutually beneficial interactions (such as potential friendship on social media) that would have taken much longer to form serendipitously. Numerous methods have been proposed to solve this classical problem. Yet, existing methods are typically designed for undirected networks, and their performances differ greatly for networks from different domains. Here, by representing the adjacency matrices of networks as binary images and leveraging the power of deep generative models in computer vision, we developed a new link prediction method, which works for general directed or undirected complex networks. We applied this method to various real networks, finding that overall it shows superior performance against existing methods.}, URL = {https://www.biorxiv.org/content/early/2018/01/14/247577}, eprint = {https://www.biorxiv.org/content/early/2018/01/14/247577.full.pdf}, journal = {bioRxiv} }