一种新的最大间隔分类器的求解算法及其在测井岩性识别中的应用

A Novel Algorithm for Maximal Margin Classifier and Its Application in Logging Lithologic Identification

首先介绍了尺度不变凸壳的概念,提出当2个凸壳独立时,可以通过求解2个凸壳的最大间隔来建立一种新的分类器。然后通过论证尺度不变凸壳与最小闭球的关系,提出了利用最小闭球求解分类器的快速算法。通过对测井岩性识别的试验表明,与已有的方法相比,在取得相同分类成功率的同时,该算法的核估计次数和训练时间明显减少。

By introducing the notion of'scaled convex hull'(SCH) generated by the training points,a novel classifier could be constructed by maximizing the margin between two SCH when they were independent.Then,a fast algorithm to solve the classifier is presented by building the relationship between the SCH and the minimum enclosing ball.The experiments on the data of logging lithological character identification show that the proposed method may achieve better performance than the state-of-the-art methods,in terms of kernel evaluations and execution time.