Research And Implementation Of Genetic Algorithm Based BN Augmented Na(?)ve-Bayes Classifier

Classification has always been a central issue on data mining, machine learning and pattern recognition, classifier, as an important model and method of machine learning and data mining, is very important to the development and application of machine learning and the data mining.The classifier's effect closely correlates with the characteristic of data sets, at present,the construction of classifier is generally based on the character of different datasets, there is no such a classifier which is suitable for any data sets.Under uncertain conditions, the Bayes network is a powerful tools for the knowledge expression and inference, but for difficulties in constructing its network structure and very high time complexity, it has not been considered as a classifier algorithm until the emergence of Na?ve-Bayes Classifier.Under the condition independent assumption , Na?ve-Bayes Classifier has a very high study efficiency and accuracy, but this kind of assumption is not always realistic , it shows low accuracy when the attribute variables in a data set strongly correlated with each other. The TAN relaxed the condition independent hypothesis , by allowing the attributes variable to form a tree which represents their inferior correlation , it is proved to be highly effective and accurate.However, when there are more attribute variables that have complex correlations with each other, the tree structure are unable to reflects the real relations between the attributes , as a result , its accuracy drops down. The BAN further relaxed the condition independent assumption, allowing the attributes to from a Bayes network.For Bayes network is a powerful tools for knowledge expression and inference, BAN can effectively reflect the intrinsic relations between the attributes. However, because Bayes network structure learning algorithm generally explore too big a search space , as well as its computing complexity, and moreover, the such learning algorithm is very easy to fall into the local optimal solution, BAN's efficiency is questioned .Finding a kind of highly effective global searching algorithm is an extremely important to the study of BAN.Genetic algorithm are a kind of biological modelling algorithm to solve the global optimization question, which takes Darwin's theory of nature evolution and Mendel's theory of hereditary and mutation as its foundation. Genetic algorithm is...