MSDA-NMF: A Multilayer Complex System Model Integrating Deep Autoencoder and NMF

In essence, the network is a way of encoding information underlying social management system. Ubiquitous systems rarely exist alone and have dynamic complexity. For complex systems, it difficult to extract represent multi-angle features data only by using non-negative matrix factorization. Existing deep NMF models integrating multi-layer struggle explain results obtained after mid-layer NMF. this paper, introduced into structure, feature representation input realized hierarchical structure. By adding regularization constraints for each layer, essential are characterizing transformation layer-by-layer. Furthermore, autoencoder fused construct model MSDA-NMF that integrates autoencoder. Through multiple sets such as HEP-TH, OAG Pol blog, Orkut Livejournal, compared with 8 popular models, Micro index better increased 1.83, NMI value 12%, link prediction performance improved 13%. robustness proposed verified.