用于微阵列数据分类的子空间融合演化超网络

摘要

针对传统模式识别方法在学习具有小样本特性的DNA微阵列数据时存在的过拟合问题，本文提出了一种子空间融合演化超网络模型。该模型通过子空间划分、超边全覆盖和子空间融合三种方法降低模型对初始化的依赖，减少了对数据空间的拟合误差，提高了演化超网络的泛化能力。对四个DNA微阵列数据集的实验结果表明，子空间融合演化超网络的识别率和在小样本训练集下的泛化能力均优于参与对比的其他传统模式识别方法。%In order to solve the over-fitting problem of the traditional pattern recognition approaches under the DNA microarray data with small train samples,a subspace fusion-based evolutionary hypernetwork model is proposed in this pa-per.With the methods of subspace division,hyperedge coverage,and subspace fusion,the proposed scheme reduces the de-pendence on the initialization,decreases the fitting error of the data space,and enhances the generalization ability of the evo-lutionary hypernetwork.The experimental results on four DNA microarray datasets show that the proposed model achieves higher classification accuracy and stronger generalization ability than other compared traditional pattern recognition method.