文献类型：专利

英文题名：Matrix classification method based on interclass discrimination, involves classifying unknown matrix pattern according to output result

作者：WANG Z;LI D;ZHANG G;GAO D

机构：[1]UNIV EAST CHINA SCI & TECHNOLOGY

申请号：CN107025461-A

申请日：2016-12-08

公开日：2017-08-08

语种：英文

收录：DERWENT

摘要：NOVELTY - The method involves collecting the samples into matrix mode so that the algorithm is processed. A new matrix model CBCMatMHKS is generated, and the CBCMatMHKS is trained with a training set. An optimal solution of CBCMatMHKS is obtained by a gradient descent method. The test set is used to test the optimal solution and get an optimal decision function. An optimal decision function is used to calculate a matrix pattern of an unknown class of input, and the unknown matrix pattern is classified according to the output result. USE - Matrix classification method based on interclass discrimination for use in pattern recognition. ADVANTAGE - The method maximizes the distance between the different types of samples, by introducing the discriminant information between classes, using the cluster heart to represent the samples of a certain region. The classification correct rate is improved. DESCRIPTION OF DRAWING(S) - The drawing shows a flowchart of the matrix classification method. (Drawing includes non-English text)