Methods of Pattern Classification

JIANG Bin,LI Xiang,WANG Hong-qiang,GUO Gui-rong
DOI: https://doi.org/10.3321/j.issn:1001-506x.2007.01.025
2007-01-01
Abstract:Pattern classification is not only an important aspect in pattern recognition but also the key technique in transaction of other questions.Firstly the method of FKNN is analyzed.Secondly the method of FMMNN is also studied.Thirdly studying the theory of SVM,this paper presents an ameliorated RBF kernel function.Finally,the simulation is processed based on the IRIS data and the results are compared with each other.These results show that the performance of the ameliorated SVM outperforms those of the FMMNN and FKNN,with shorter operation time and more suitable for real-time implementation.
What problem does this paper attempt to address?