中文摘要：

马田系统是多变量数据挖掘中模式识别方法的新进展,变量间的复共线性会通过马氏距离影响马田系统变量筛选的效果和判别的准确率。为了克服复共线性对马田系统的负面影响,提出了基于岭估计新的测量尺度—岭马氏距离,通过变量敏感性和条件数绘制三条岭迹来确定岭参数,并设计了自适应多目标遗传算法进行基准空间优化,使得马田系统分类方法对病态数据具有更好的耐受性。通过案例验证了岭马氏距离可以很好的克服复共线性,并提高马田系统分类方法的判别准确率。

英文摘要：

MTS is a new progress of pattern recognition in the area of multi-dimensional data mining,in which the performance of variable screening and discrimination accuracy will be affected by multicollinearity among variables through MD.To overcome the negative effect of multicollinearity to MTS,a new measuring scale RMD based on ridge estimation is presented,three ridge traces on the sensitiveness and condition number are drawn to determine the ridge parameter,and an algorithm AMOGA is designed for optimizing the Mahalanobis space.These measures make MTS more robust to data with multicollinearity.A case study shows that RMD has a good capability to overcome multicollinearity and improve the accuracy of MTS.