Investigation of a framework based on artificial immune system applied to the pattern recognition / InvestigaÃÃo de uma arquitetura baseada em sistemas imunolÃgicos artificiais aplicada ao reconhecimento de dÃgitos manuscritos

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

The discovery of new functionalities through the study of human physiology has contributed toward the evolution of Artificial Immune Systems. The present work investigates a new architecture through observations of natural immunological behavior. New functionalities observed in the biological environment were studied for the modeling of this immunological approach, principally the organization and clustering of similar antibodies (Ab) throughout the process. It is believed that these functionalities may improve the recognition capacity of artificial immune algorithms. This technology tries to model defense mechanism characteristics and functionalities of living beings. The defense mechanism allows an organism to defend against invasion from foreign substances. The natural immune system stores the best antibodies in its genetic memory. These are later used to identify antigens that have previously invaded the organism, thereby obtaining a quicker, more efficient response. The performance of this model was compared to other techniques found in the literature in order to assess the response quality of the algorithm. MLP (Multi-Layer Perceptron), k-NN (k-Nearest Neighbor) and Clonalg (Clonal Selection Algorithm) were used. Thus, it was possible to estimate the contribution this new model may have in research regarding Artificial Immune Systems and pattern recognition

ASSUNTO(S)

reconhecimento de dÃgitos manuscritos ciencia da computacao sistemas imunolÃgicos artificiais sistemas hÃbridos inteligentes hybrid intelligent systems artificial immune system handwritten digits recognition

ACESSO AO ARTIGO

Documentos Relacionados