Features of edge-centric collective dynamics in machine learning tasks
Paulo Roberto Urio1; Filipe Alves Neto Verri2; Liang Zhao3
1 Institute of Mathematical and Computer Sciences, University of São Paulo; 2 School of Electrical, Computer and Energy Engineering, Arizona State University; 3 Ribeirão Preto School of Philosophy, Science and Literature, University of São Paulo
Resumo
We study how collective dynamics can solve Machine Learning tasks. The Edge Domination System is an algorithm to reveal patterns and obtain information of the underlying complex network. The algorithm consists in the simulation of a dynamical system based on particle competition for the dominance of edges. In this paper, we apply this method to semi-supervised tasks of data learning problems. We propose a vertex-centric version of this model and assess the differences between the edge-centric model.
Palavras-chave: Analysis and Control of Nonlinear Dynamical Systems with Practical Applications; Complex Networks; Nonlinear Dynamics and Complex Systems; Collective Dynamics and Particle Competition; Machine Learning