Machine Learning in Complex Networks
This book intends to combine two widely studied research areas, machine learning and complex networks, which in turn may generate broad interests to scientific community, mainly to computer science and engineering areas
Springer International Publishing
|Edition:||1st ed. 2016|
|Collection:||Springer eBooks 2005- - Collection details see MPG.ReNa|
|Summary:||This book intends to combine two widely studied research areas, machine learning and complex networks, which in turn may generate broad interests to scientific community, mainly to computer science and engineering areas|
This book explores the features and advantages offered by complex networks in the domain of machine learning. In the first part of the book, we present an overview on complex networks and machine learning. Then, we provide a comprehensive description on network-based machine learning. In addition, we also address the important network construction issue. In the second part of the book, we describe some techniques for supervised, unsupervised, and semi-supervised learning that rely on complex networks to perform the learning process. Particularly, we thoroughly investigate a particle competition technique for both unsupervised and semi-supervised learning that is modeled using a stochastic nonlinear dynamical system. Moreover, we supply an analytical analysis of the model, which enables one to predict the behavior of the proposed technique. In addition, we deal with data reliability issues or imperfect data in semi-supervised learning.
Even though with relevant practical importance, little research is found about this topic in the literature. In order to validate these techniques, we employ broadly accepted real-world and artificial data sets. Regarding network-based supervised learning, we present a hybrid data classification technique that combines both low and high orders of learning. The low-level term can be implemented by any traditional classification technique, while the high-level term is realized by the extraction of topological features of the underlying network constructed from the input data. Thus, the former classifies test instances according to their physical features, while the latter measures the compliance of test instances with the pattern formation of the data. We show that the high-level technique can realize classification according to the semantic meaning of the data.
|Physical Description:||XVIII, 331 p. 87 illus., 80 illus. in color online resource|