文章摘要
基于数据处理分组法(GMDH)的遗传选择算法与数据挖掘克隆技术
  
DOI:
中文关键词: 数据处理分组法的多层算法  遗传算法  遗传选择
英文关键词: GMDH MIA algorithm  genetic algorithm  genetic selection
基金项目:河南省自然科学基金项目(122102210510)
作者单位
魏雪峰,孙利 黄淮学院 信息工程学院,河南 驻马店 463000 
摘要点击次数: 1257
全文下载次数: 170
中文摘要:
      数据处理分组法(GMDH)的多层算法(MIA)是通过利用遗传算法的选择程序来实现的,该算法包含了为获取更少误差而生成的最优神经元的克隆操作.选择程序根据适应度以及网络输入的某种概率从已有的神经元中为新神经元寻找父代.克隆实质是对最优神经元的克隆体参数进行稍微调整.从机器学习库选出实验结果表明,通过克隆操作进行基因改造后的GMDH网络表现得比其它方法优越.
英文摘要:
      Grouping method of data processing (GMDH) the multilayered algorithm (MIA) was achieved through the use of genetic algorithm selection procedures, the algorithm get less errors generated optimal neuron cloning operation. The procedure was selected based on the probability of fitness, as well as some of the network input that from existing neurons parent looking for new neurons. Essentially cloning the optimal neurons clones parameters slightly adjusted. Experimental results show that genetically modified by cloning operation the GMDH network performance is superior other methods that elected from the machine learning library.
查看全文   查看/发表评论  下载PDF阅读器
关闭