Handwritten Data Clustering Using Agents Competition in Networks

Machine learning in complex networks: modeling, analysis, and applications

“…The local clustering coefficient of a vertex in a graph quantifies how close its neighbors are to being a clique (complete graph). Mathematically speaking, the local clustering coefficient of vertex i that is member of the component j is given by: 6) where |e uk | the number of links shared by the direct neighbors of vertex i (number of triangles formed by vertex i and any of its two neighbors) and k (j) i is the degree of vertex i of component j. By (2.6), we see that CC…”

“…In brief, the step responsible for defining the cliques is the most time-consuming one. 6 C is the number of communities in the network. The critical point of this algorithm is the evaluation of the Q-fitness function.…”

Machine learning via dynamical processes on complex networks

“…As a consequence, the application of network-based methods in learning tasks has been increasing over the past years and has become a active research area with a myriad of applications such as semi-supervised learning (Breve et al, 2012;Chapelle et al, 2006;Nguyen and Mamitsuka, 2011;Silva and Zhao, 2012a), clustering (Karypis et al, 1999;Schaeffer, 2007;Silva and Zhao, 2012b;Silva et al, 2013), regression (Ni et al, 2012), feature selection (Bunke and Riesen, 2011), dimensionality reduction (Riesen and Bunke, 2009), among others.…”

“…Eventually, each particle will dominate a single community. In (Silva et al, 2013), the authors applied the same concept of agents competition to cluster handwritten images of alphanumeric digits. Another method of community detection (Zhou, 2003a,b) uses a distance measure of complex networks based on the random walk of a Brownian particle through the network.…”

“…In (Quiles et al, 2008;Silva and Zhao, 2012a,b;Silva et al, 2013), random walk is used in combination with preferential walk in competitive processes. Competing particles walk through nodes in a network and the number of passages of a particle through a node determines its domination over that node.…”