A non-linear approach for pattern recognition in networks

Abstract can also be found here (page 249) The combination of pattern recognition and networks emerges as an important approach to the high demand for methods that handle in a big data scenario. Thus, pattern recognition in networks aims to characterize networks by extracting information from the correlation between vertices and their relationship to the network topology. While studies have tended to focus on structural measurements, which are commonly used for the characterization of networks (1), recently, there have been some attempts to use non-linear methods, such as random walks. (2) Thus, we proposed a cellular automata model that embeds its dynamics over a network topology aiming to produce their intrinsic spatio-temporal patterns from them. Inspired by the rules of an artificial life model, here we introduce the life-like network automata (LLNA) (3) as a tool for network analysis in the context of pattern recognition applications. The LLNA method performances was explored into four classification tasks that use networks as data representation for pattern recognition purposes: (i) to infer the authorship of disputed documents in multiple contexts; (ii) identifying organisms from distinct domains of life, Archaea, Bacteria and Eukaryota, through their metabolic networks; (iii) identifying structural patterns in two online social networks, Twitter and Google+, and, finally, (iv) classifying stomata distribution patterns varying according to different lighting conditions. The proposed method has made significant progress in real-world pattern recognition applications from a wide branch of fields. 1 COSTA, L. F. et al. A pattern recognition approach to complex networks. Journal of Statistical Mechanics: theory and experiment, v. 2010, p.1015-1-11015-24, 2010. doi: 10.1088/1742-5468/2010/11/P11015. 2 GONÇALVES, W. N.; MARTINEZ, A. S.; BRUNO, O. M. Complex network classification using partially self-avoiding deterministic walks. Chaos, v. 22, n. 3, p. 033139-1-033139-13 , 2012. 3 MIRANDA, G. H. B.; MACHICAO, J.; BRUNO, O. M. Exploring spatio-temporal dynamics of cellular automata for pattern recognition in networks. Scientific Reports, v. 6, p. 37329-1-37329-15, 2016. doi: 10.1038/srep37329.Presentation given during the Semana integrada da graduação e pós-graduação do Instituto de Física de São Carlos, 2017, São Carlos, Brasil