论文部分内容阅读
特征选择已经是高维数据处理尤其是模式识别领域中的一个关键问题.提出一种混合特征选择模型用于从潜在的相关特征中选择那些最重要的特征.该模型包括两部分:filter部分与wrapper部分.在filter部分,4种不同的Filter方法分别对候选特征进行独立排序,在融合后进一步生成综合特征排序,综合排序随后产生遗传算法(GA)的初始种群.在wrapper部分,GA算法根据神经网络的分类准确率对个体(特征子集)进行评价,以便于搜索到最优的特征子集.测试结果表明,该模型不仅能有效地减少特征子集的大小,而且还可以进一步提高分类识别的准确率和效果.
Feature selection has been a key issue in the field of high-dimensional data processing, especially in pattern recognition. A hybrid feature selection model is proposed to select the most important features from the potential related features. The model consists of two parts: wrapper. In the filter part, four different Filter methods respectively separate the candidate features separately, and then generate a comprehensive feature order after fusion, and then generate an initial population of genetic algorithm (GA) The classification accuracy of neural networks can be used to evaluate individuals (feature subsets) in order to search for the best subset of features.The test results show that this model not only can effectively reduce the size of feature subsets, but also can further improve the classification Recognition accuracy and effectiveness.