Smooth Deep Network Embedding

摘要: Network embedding is an efficient method to learn low-dimensional representations of vertexes in networks since the network structure can be captured and preserved through this process. Unlike shallow models, deep neural framework able capture highly non-linear structure. Therefore, it achieve much better performance comparison traditional methods. However, few attention has been paid smoothness such contrast numerous research works for image text fields. Methods without are not robust enough, which means that slight changes on may lead dramatic results. Hence, how find a smooth still open yet important problem. To end, paper, we propose Smooth Deep Embedding method, namely SmNE, generates stable reliable Empirically, conduct experiments real-world networks. The results show compared state-of-the-art methods, our proposed significant gains several applications.