Um estudo sobre reconhecimento de padrões: um aprendizado supervisionado com classificador bayesiano / A study on pattern recognition: supervised learning with a Bayesian classier

AUTOR(ES)

Pedro Henrique Ramos Cerqueira

DATA DE PUBLICAÇÃO

2011

RESUMO

The facility we have to recognize a face, to understand spoken words, reading manuscripts, identifying car keys in our pocket and deciding whether an apple is ripe by its smell, belie the complex processes that are behind the act to recognize these patterns. These recognitions have been crucial to our survival, and over the past tens of millions of years sophisticated systems were developed to accomplish these tasks. The pattern recognition has aimed to carry out the rating of a given set of data in certain classes or groups, considering their standards and those of their classes, allowing several applications, such as: document processing, bar code readers, identication of people, optical and ngerprint drivers, industrial automation, image processing and agronomic applications, molecular markers analysis and plants classication, which became a technique of great importance in recent years. For a better rating is necessary to perform some studies, which can be formulated by supervised and unsupervised methods, in order to develop classiers, such as the Bayesian classier and neural networks, which enable the decision-making. To check the quality of ratings, specic measurements must be used, such as the kappa index or the general error probability. Thus it is essential to use software to make a decision, including the R and WEKA, developed specically to solve pattern recognition problems. In order to solve problems of specic areas, as automation and agronomy, the supervised learning method with Bayesian classier was used, and to verify the ratings quality, the kappa index and the general error probability were used. Software R and WEKA were used, in order to perform ratings for molecular markers data, soybean data and types of packaging for auto parts.

ASSUNTO(S)

aprendizagem classication classificação estatística computacional inferência bayesiana learning marcador molecular mineração de dados molecular marker pattern recognition probabilidade reconhecimento de padrões softwares. statistical nnnnadaaaaacomputing.

Documentos Relacionados

• BAYESIAN LEARNING FOR NEURAL NETWORKS
• Análise não-linear no reconhecimento de padrões sonoros : estudo de caso para sons pulmonares
• "Data pre-processing for supervised machine learning"
• Algoritmo de aprendizado supervisionado - baseado em máquinas de vetores de suporte - uma contribuição para o reconhecimento de dados desbalanceados
• Pattern recognition in integrated sensors.