Study On The Reasoning Under Uncertainty And Data Analysis Oriented Pattern Recognition Methods

In many areas, such as Artificial Intelligence, Machine Learning, Pattern Recognition and data mining, there always exist imprecise and uncertainty in intelligent systems, so we must deal with problems under uncertainty.For these who research into image processing and pattern recognition, the problems under consideration can be seen as reasoning under uncertainty. So, we study the pattern recognition problems from the view of uncertainty reasoning and reveal the nature of various pattern recognition methods, which will be significant.Probability theory is the only reasonable way to represent uncertainty, it is useful for machine learning or reasoning under uncertainty. Bayesian probability theory is a branch of mathematical probability theory that allows one to model uncertainty about the world and outcomes of interest by combining common-sense knowledge and observational evidence. Therefore, we would better unify various machine learning and pattern recognition problems into the framework of Bayesian machine learning method, and consider them as a Bayesian inference problem: that is, all of machine learning can be phrased in terms of computing the posterior over the parameters given fully observed data. This provides a conceptually clean and logically coherent"story"that ties all the material together.Bayesian networks are the combination of probability theory and graph theory, representing the conditional independences among variables and their probability distribution. They are used for probability inference and are efficient methods adapted for reasoning under uncertainty. One of the strongest reasons is that uncertainty is indeed prevalent in the real world and probability theory is the reasonable way to represent uncertainty, so Bayesian networks are critical to represent that uncertainty in at least a semi-coherent manner. Another reason is that Bayesian networks are a fundamentally more modular representation of uncertain knowledge. This makes them easier to maintain, and to adapt them to different contexts. So, Bayesian networks are quite attractive for reasoning under uncertainty.