基于代价敏感大间隔分布机的不平衡数据分类算法华东理工大学学报.PDFVIP

华东理工大学学报（自然科学版） Journal of East China University of Science and Technology (Natural Science Edition) DOI：10.14135/ki.1006 -3080. 20180515001 基于代价敏感大间隔分布机的不平衡数据分类算法1 曹雅茜，黄海燕 （华东理工大学化工过程先进控制和优化技术教育部重点实验室，上海200237 ） 摘 要：大间隔分布学习机(LDM)在应用于不平衡据分类时，由于忽略类别不均衡，会使少数 类样本的识别率较低。针对这一不足，结合代价敏感思想提出了一种不平衡代价敏感大间隔分布 算法(ICS-LDM) 。首先，在计算间隔均值和间隔方差时，结合数据集的不平衡因子和样本错分代 价参数，调整不同类别的间隔分布权重；其次，将可以快速收敛的循环对偶坐标下降法应用于求 解目标函数；最后，通过逐渐提高少数类的间隔分布，可以实现间隔分布在各类别平衡且总体最 大。在虚拟数据集和UCI 公开数据集上的实验结果表明，ICS-LDM 可以有效提高少数类的分类 精度，平衡各类的分类性能。 关键词：不平衡数据；代价敏感学习；大间隔分布；循环对偶坐标下降法 中图分类号：TP391.9 文献标志码：A Imbalanced Data Classification Based on Cost-Sensitive Large Margin Distribution Machine CAO Ya-xi, H UANG Hai-yan (Key Laboratory of Advanced Control and Optimization for Chemical Processes , Ministry of Education, East China University of Science and Technology, Shanghai 200237, China) Abstract: The recent theoretical revealed that compared with a single margin, margin distribution is more critical to the generalization performance. Large margin distribution machine (LDM) is put forward to get superior classification performance and strong generalization performance by maximizing the margin mean and minimizing the margin variance simultaneously. However, classifiers tend to be overwhelmed by the majority classes and ignore the minority class because of class imbalanced. Therefore, LDM easily leads to the lower detection rate of the minority class on imbalanced data, which contradicts to the needs of high detection rate of the minority class in many real applications. In this paper, imbalanced cost-sensitive large marg