中文摘要：

为提高含有异常值数据集的学习性能,对基于支持向量机的鲁棒算法进行了研究,深入分析了异常值降低标准支持向量机推广能力的本质原因,从基于支持向量机的异常值检测和抑制异常值对支持向量机的影响两个方面,较为系统地回顾了国内外在该领域的研究发展现状和最新研究进展,其中包括各种算法的基本思想和主要特点。归纳总结了支持向量机关于异常值问题的主要研究内容、方法、研究成果以及存在的问题,并进一步提出了在应用方面的研究方向。

英文摘要：

To improve the performance of the support vector machine （SVM） to the presence of outlier samples, the robust approaches based on SVM are studied. First, the internal reason why the generalization performance of SVM is deteriorated by outliers is analyzed. Then the current research trends and the latest developments of SVM on detecting and suppressing outliers are reviewed, which introduce the basic ideas of the approaches and their major characteristics. The aim of the paper is to sum up the studies on outliers based on SVM, including issues, methods and results. Finally, the new research prospects of this domain are also proposed.