Research On Pattern Recognition Method Based On Feature Primitives Of Graphical Representation

In the research of pattern recognition, we always face with multi-dimension data which can indicate lots of detailed information. Researchers paid unremitting efforts on how to denote these data directly and explore interesting information from the data. On base of pioneer's achievements, the method of pattern recognition method based on feature primitives of multi-graph is researched in this paper.The paper's background was the issue of representation and classification of data in pattern recognition. A novel method for feature primitives extraction of multivariate graph is introduced, which is based on 2D graphical representation theory of multivariate data.First, the methods of multi-dimensional data visualization coupled with the theory of multivariate graphical representation are discussed thoroughly. The method of representation (geometric representation) was researched. Based on the model of feature primitives of multivariate graph information, the graph feature primitives is extracted. Then, the feature order issue is occurred in the course of the multivariate graphical representation, and shall result in that the classification performance should be different based on the generated graphical feature of the multivariate graph of different feature order. So the corresponding feature order methods were researched. The first method was the feature order of filter. The second method was the feature order based on wrapper. The Third method was Genetic Algorithm. Some data experiments using the Iris dataset of the UCI repository of machine learning databases are accomplished, and it is shown that this method achieves better recognition result. By using SVM learning method for dataset classification, we investigate the classification error of different parameter combinations and analyze the parameters'impact on classification accuracy. Finally, the paper summed up the main work and analyze the existing problems, provide a basis and valuable experience for further study.