论文部分内容阅读
针对模式分类问题,提出一种具有磁场效应的p-间隔核学习机(p-MKLM),旨在寻求一个具有磁场效应的最优超平面,受其吸引,使得一类模式离该平面的距离尽可能的小,而另一类模式受其排斥,离该平面的间隔尽可能的大,从而最大可能地实现模式分类.通过引入一个可调节的磁场强度q,减小一类模式的数据分布,从而提升分类性能.分别采用人工数据和实际数据进行实验,所得结果显示,p-MKLM在一类和二类模式分类上的性能均优于或等同于相关方法.
Aiming at the problem of pattern classification, a p-space kernel learning machine (p-MKLM) with magnetic field effect is proposed to find an optimal hyperplane with magnetic field effect, which is attracted by it. The p- As small as possible while the other modes are rejected by them as much as possible from the plane to achieve the maximum possible pattern classification. By introducing an adjustable magnetic field strength q, the data distribution of a class of patterns is reduced , So as to improve the classification performance.Experimental results using artificial data and real data respectively show that the performance of p-MKLM is better than or the same as the related methods in the classification of the first and second modes.