一种基于超长方体集的模糊模式识别算法

Algorithm for Fuzzy Pattern Recognition Using Hypercube Set

结合模糊C均值算法(FCM)与模糊最小最大神经网络算法,提出一种基于超长方体集的模糊模式识别算法.首先采用基于特征加权的FCM算法进行粗划分,得到c个平行于特征轴的超椭圆球类;再根据已知的样本的类别标记进行进一步划分;以改进的最小最大模型建立超长方体阵,使每个超长方体只能容纳一种类别的样本点,并且分属不同类别的超长方体无重叠.使用这种方法构建超长方体集更快速、更简单,但对训练样本的要求较高,应选择能够充分体现数据分布情况的数据点作为训练样本.

Based on Fuzzy C-Means algorithm (FCM) and Fuzzy Min-Max Neural Networks,an integrated algorithm for fuzzy pattern recognition using hypercube set was proposed.The (training) dataset was classified into subsets using FCM algorithm based on feature weighting.Further classification was performed on the subsets based on the known classification marks of the training set.A hypercube set was created using modified Fuzzy Min-Max Neural Networks and met the following criteria:I) Each hypercube only holds the data points with the same classification mark.II) There is no overlap of the hypercubes that belong to different classes.This algorithm can speed up and simplify the creation of the hypercube set.However,it requires a careful selection of the training sample.The sample should fully represent the pattern distribution of the population.