一种KD树集成偏标记学习算法  

A ensemble K-dimension tree for partial label learning

作　　者：卢勇全 刘振丙[1,2] 颜振翔 方旭升 LU Yongquan;LIU Zhenbing;YAN Zhenxiang;FANG Xusheng(School of Electronic Engineering and Automation,Guilin University of ElectronicTechnology,Guilin 541004,China;School of Computer and Information Security,Guilin University of ElectronicTechnology,Guilin 541004,China)

机构地区：[1]桂林电子科技大学电子工程与自动化学院,广西桂林541004 [2]桂林电子科技大学计算机与信息安全学院,广西桂林541004

出　　处：《桂林电子科技大学学报》2019年第6期454-459,共6页Journal of Guilin University of Electronic Technology

基　　金：国家自然科学基金(61562013,61866009);广西自然科学基金(2017GXNFDA198025);桂林电子科技大学研究生教育创新计划(2017YJCX101)。

摘　　要：针对样本集不均衡造成分类器精度不足的问题,提出一种KD树均衡训练集的集成偏标记学习算法。按照伪标签划分样本,采用KD树检索的方式均衡训练集,再采用多个分类器投票方式实现消岐,最终运用集成学习的方法实现分类。在公开数据集上的仿真实验结果表明,该偏标记学习算法在分类上具有较好的表现力。In partial label learning framework,how to eliminate the impact of candidate label on model learning is crucial.Aiming at the lack of accuracy of the classifier due to the imbalance of the sample set,an ensemble partial label learning method for K-dimension tree equilibrium training set is proposed in this paper.Firstly,the samples are divided according to the candidate label,and the training set is balanced by the KD tree retrieval method.Then,multiple classifier voting methods are used to achieve the elimination.Finally,the ensemble learning method is used to realize the classification.The simulation experiments are carried out on the public dataset.The experimental results show that the proposed partial marker learning algorithm has better performance for classification.

关 键 词：偏标记学习 伪标签 KD树 集成学习 均衡训练集 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]