采用概率密度比值估计的距离度量学习  

Distance metric learning with Probability Density Ratio Estimation

作　　者：吕清秀[1] 李弼程[1] 高毫林[1] 

机构地区：[1]解放军信息工程大学信息系统工程学院,郑州450002

出　　处：《信号处理》2013年第5期607-614,共8页Journal of Signal Processing

基　　金：全军军事学研究生课题资助项目(YJS1062)

摘　　要：现有的距离度量学习算法都是假设训练数据和测试数据服从相同的分布,但是该假设在实际中不一定成立。当训练数据和测试数据的分布不同时,利用训练数据学习得到的度量函数可能难以适用于测试数据。针对上述问题,本文在NCA(Neighbourhood Components Analysis)度量学习方法的基础上,通过引入概率密度比值对目标函数加权,提出了一种采用概率密度比值估计的距离度量学习方法(Distance metric learning with ProbabilityDensity Ratio Estimation,DML-PDR)。在UCI数据集和Corel图像库上的KNN分类实验表明,新方法克服了传统度量学习方法的不一致问题,提高了分类的准确率。Previous distance metric learning algorithms assume that the training data and test data have the same distri- bution, but the assumption may be not always true in practice. When the training data and test data have different distribu- tion, the distance metric learned from the training data may be not fit for test data. In order to resolve above-mentioned problem, this paper propose a novel distance metric learning with probability density ratio estimation based on NCA（ Neigh- bourhood Components Analysis）, which weight the objective function by applying the probability density ratio. The KNN classification on UCI data sets and Corel images demonstrate that the new method resolve the inconsistent of traditional dis- tance metric learning.

关 键 词：距离度量学习 半正定规划 概率密度比值估计 图像分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]