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.
作者
卢勇全
刘振丙
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)
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)。
关键词
偏标记学习
伪标签
KD树
集成学习
均衡训练集
partial label learning
candidate label
K-dimension tree
ensemble learning
balance training set
