一种多标签随机均衡采样算法
为解决多标签学习中数据不平衡、传统重采样过程标签样本集相互影响以及弱势类信息大量重复和强势类信息大量丢失的问题，提出多标签随机均衡采样算法。该算法在多标签的条件下提出随机均衡采样思想，充分利用强势类和弱势类信息来平衡数据冗余和损失；优化样本复制和删除策略，保证不同标签重采样过程的独立性；提出平均样本数，保持数据的原始分布。实验在三个数据集下对比了三种多标签重采样算法的性能，结果表明，0．2和0．25是所提算法的最佳重采样率，且该算法尤其适用于不平衡度较高的数据集，与其他方法相比具有最好的性能。...
Saved in:
| Published in | 计算机应用研究 Vol. 34; no. 10; pp. 2929 - 2932 |
|---|---|
| Main Author | |
| Format | Journal Article |
| Language | Chinese |
| Published |
国家数字交换系统工程技术研究中心,郑州,450002
2017
|
| Subjects | |
| Online Access | Get full text |
| ISSN | 1001-3695 |
| DOI | 10.3969/j.issn.1001-3695.2017.10.011 |
Cover
| Summary: | 为解决多标签学习中数据不平衡、传统重采样过程标签样本集相互影响以及弱势类信息大量重复和强势类信息大量丢失的问题，提出多标签随机均衡采样算法。该算法在多标签的条件下提出随机均衡采样思想，充分利用强势类和弱势类信息来平衡数据冗余和损失；优化样本复制和删除策略，保证不同标签重采样过程的独立性；提出平均样本数，保持数据的原始分布。实验在三个数据集下对比了三种多标签重采样算法的性能，结果表明，0．2和0．25是所提算法的最佳重采样率，且该算法尤其适用于不平衡度较高的数据集，与其他方法相比具有最好的性能。 |
|---|---|
| Bibliography: | 51-1196/TP To deal with the class imbalance in multi-label learning, the interaction between different label sets, the information redundancy of minority classes as well as the loss of majority classes that existed in traditional multi-label resampling, this paper put forward a multi-label random balanced resampling algorithm. The algorithm proposed the random balanced resampling method to make use of minority and majority information to balance the redundancy and the loss, improved replication and deletion stra- tegy to ensure the independence of the resampling process of different labels, and maintained the original distribution of dataset with newly proposed mean instance size. Experiment results show that the proposed method is especially suit for datasets with higher imbalance ratio and achieves the best performance, 0.2 and 0.25 are the best resampling ratios in it. Li Sihao, Chen Fucai, Huang Ruiyang （ National Digital Switching System Engineering ＆ Technological R＆D Center, Zhengzhou 450002, China） multi |
| ISSN: | 1001-3695 |
| DOI: | 10.3969/j.issn.1001-3695.2017.10.011 |