Deep Auto-encoded Clustering Algorithm for Community Detection in Complex Networks

• Published in: Chinese Journal of Electronics Vol. 28; no. 3; pp. 489 - 496
• Main Authors: Wang, Feifan, Zhang, Baihai, Chai, Senchun
• Format: Journal Article
• Language: English
• Published: Published by the IET on behalf of the CIE 01.05.2019
• Subjects: Autoencoder; clustering strategies; Community detection; community detection algorithm; Complex networks; DAC; deep auto‐encoded clustering algorithm; Deep learning; eigenvalues and eigenfunctions; graph theory; key community information; k‐means community detection method; learning (artificial intelligence); lower‐dimensional representation; matrix algebra; network theory (graphs); pattern clustering; real world networks; sparse single autoencoders; sparsity constraints; spectral community detection algorithm; synthetic world networks; time‐consuming eigenvalue decomposition procedure; unsupervised autoencoders; lower-dimensional representation; spectral community detection algorithm; sparse single autoencoders; deep auto-encoded clustering algorithm; graph theory; DAC; network theory (graphs); Community detection; complex networks; synthetic world networks; key community information; k-means community detection method; eigenvalues and eigenfunctions; matrix algebra; deep learning; community detection algorithm; sparsity constraints; pattern clustering; Autoencoder; learning (artificial intelligence); real world networks; time-consuming eigenvalue decomposition procedure; clustering strategies; unsupervised autoencoders
• ISSN: 1022-4653; 2075-5597
• DOI: 10.1049/cje.2019.03.019

The prevalence of deep learning has inspired innovations in numerous research fields including community detection, a cornerstone in the advancement of complex networks. We propose a novel community detection algorithm called the Deep auto-encoded clustering algorithm (DAC), in which unsupervised and sparse single autoencoders are trained and piled up one after another to embed key community information in a lower-dimensional representation, such that it can be handled easier by clustering strategies. Extensive comparison tests undertaken on synthetic and real world networks reveal two advantages of the proposed algorithm: on the one hand, DAC shows higher precision than the k-means community detection method benefiting from the integration of sparsity constraints. On the other hand, DAC runs much faster than the spectral community detection algorithm based on the circumvention of the time-consuming eigenvalue decomposition procedure.