Data Classification in Complex Networks via Pattern Conformation, Data Importance and Structural Optimization

  • Murillo G. Carneiro
  • Liang Zhao

Resumo


Most data classification techniques rely only on the physical features of the data (e.g., similarity, distance or distribution), which makes them difficult to detect intrinsic and semantic relations among data items, such as the pattern formation, for instance. In this thesis, it is proposed classification methods based on complex networks in order to consider not only physical features but also capture structural and dynamical properties of the data through the network representation. The proposed methods comprise concepts of pattern conformation, data importance and network structural optimization, which are related to complex networks theory, learning systems, and bioinspired optimization. Extensive experiments demonstrate the good performance of our methods when compared against representative state-of-the-art methods over a wide range of artificial and real data sets, including applications in domains such as heart disease diagnosis and semantic role labeling.

Publicado
06/07/2017
Como Citar

Selecione um Formato
CARNEIRO, Murillo G.; ZHAO, Liang. Data Classification in Complex Networks via Pattern Conformation, Data Importance and Structural Optimization. In: CONCURSO DE TESES E DISSERTAÇÕES (CTD), 30. , 2017, São Paulo. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2017 . ISSN 2763-8820. DOI: https://doi.org/10.5753/ctd.2017.3463.