基于相对贡献率的噪声裁剪算法

doi:10.16182/j.issn1004731x.joss.19-FZ0289

系统仿真学报 ›› 2019, Vol. 31 ›› Issue (12): 2721-2730.doi: 10.16182/j.issn1004731x.joss.19-FZ0289

基于相对贡献率的噪声裁剪算法

刘朔瑜, 戴月明

江南大学物联网工程学院,江苏无锡 214122

收稿日期:2019-04-16 修回日期:2019-07-07 发布日期:2019-12-13
作者简介:刘朔瑜(1993-),男,江苏无锡,硕士生,研究方向为自然语言处理; 戴月明(1968-),男,江苏常熟,硕士,副教授,硕导,研究方向为自然语言处理,人工智能。
基金资助:
国家自然科学基金(61973138)

Noise Clipping Algorithm Based on Relative Contribution Rate

Liu Shuoyu, Dai Yueming

School of Internet of Things Engineering, Jiangnan University, Wuxi 214122, China

Received:2019-04-16 Revised:2019-07-07 Published:2019-12-13

摘要/Abstract

摘要： 提出了一种基于相对贡献率的噪声裁剪算法(Class noise cutting, CNC)。通过计算得到特征对于主题的相对贡献率,利用特征区分度评分挑选对于当前主题分类最有价值的特征集,选出相应的候选类别,减少候选类别集,提高了分类准确率, 加快了分类器的响应速度。与另一种噪声裁剪算法(Eliminating class noise, ECN)比较,CNC具有更高的准确率,由于具有更精简的特征维度词典以及更优异的候选类别集使得响应速度大大加快。

关键词: 相对贡献率, 类别噪声裁剪, 层次结构分类, 特征选择

Abstract: This paper presents a class noise cutting algorithm (Class noise cutting, CNC) based on relative contribution rate. The algorithm calculates the relative contribution rate of features to the theme. The most valuable feature set is selected by using features distinguish rating. The corresponding candidate categories for each feature are selected, to reduece the candidate category set, improves the classification accuracy, and speed up the response speed of the classifier. Compared with another ECN noise cutting algorithm (Eliminating the class whose), CNC-has higher accuracy and because of its simpler feature dimension dictionary and better candidate category set, the response speed is greatly accelerated.

Key words: relative contribution rate, class noise cutting, hierarchical classification, feature selection

中图分类号:

TP278

刘朔瑜, 戴月明. 基于相对贡献率的噪声裁剪算法[J]. 系统仿真学报, 2019, 31(12): 2721-2730.

Liu Shuoyu, Dai Yueming. Noise Clipping Algorithm Based on Relative Contribution Rate[J]. Journal of System Simulation, 2019, 31(12): 2721-2730.

参考文献

[1] Du Y P, Liu J X, Ke W M, et al.Hierarchy Construction and Text Classfication Based on the Relaxation Strategy and Least Information Model[J]. Expert Systems with Applications (S0957-4174), 2018, 100: 157-164.
[2] 何力, 贾焰, 韩伟红. 大规模层次分类问题研究及其进展[J]. 计算机学报, 2012, 35(10): 2101-2115.
He Li, Jia Yan, Han Weihong.Category Candiadte Search in Large Scale Hierarchical Classification[J]. Chinese Journal of Computers, 2012, 35(10): 2101-2115.
[3] Sun A X, Lim E P.Hierarchical text classification and evaluation. Proceedings of the 2001 IEEE International Conference on Data Mining[C]. California, USA. 2001: 521-528.
[4] Liu T Y, Yang Y M, Wan H, et al.Support vector machines classification with a very large-scale taxonomy[J]. ACM SIGKDD Explorations Newsletter (S1931-0145), 2005, 7(2): 36-43.
[5] Silva-Palacios D, Ferri C.Improving Performance of Multiclass Classfications By Including Class Hierarchies[J]. Procedia Computer Science (S1877-0509), 2017, 108: 1692-1701.
[6] Luengo J, Shim S O, Alshomrani S, et al.CNC-NOS: Class Noise Cleaing By Ensemble Filtering and Noise Scoring[J]. Knowledge-Based Systems (S0950-7051), 2018, 140: 27-49.
[7] Zhuge H, He L.Automatic Maintenance of Category Hierarchy[J]. Future Generation Computer Systems (S0167-739X), 2017, 67: 1-12.
[8] Pereira R B, Plastino A, Zadrozny B, et al.Correlation Analysis of Performance Measure for Multi-label Classfication[J]. Information Processing & Management (S0306-4573), 2018, 54(3): 359-369.
[9] Kumar V, Pujari A K, Padmanabhan V, et al.Multi-label Classfication using Hierarchical Embedding[J]. Expert Systems with Applications (S0957-4174), 2018, 91: 263-269.
[10] Truică C O, Darmont J, Boicea A, et al.Benchmarking Top-k Keyword and Top-k Document Processing With T²K² and T²K²D²[J]. Future Generation Computer Systems (S0167-739X), 2018, 85: 60-75.
[11] Jiang L X, Wang S S, Li C Q, et al.Structure Extended Multinomial Naive Bayes[J]. Information Science (S0020-0255), 2016, 329: 346-356.
[12] Zhang L G, Jiang L X, Li C Qun, et al.Two Features Weighting Approachers For Naive Bayes Text Classfiers[J]. Knowledge-Based Systems (S0950-7051), 2016, 100: 137-144.
[13] Zhang D, Jiao L C, Bai X, et al.A Robust Semi-supervised SVM Via Ensemble Learning[J]. Applied Soft Computing (S1568-4946), 2018, 65: 632-643.
[14] Gaudioso M, Gorgone E, Labbé M, et al.Lagrangian Relaxation For SVM Feature Selection[J]. Computers & Operation Research (S0305-0548), 2017, 87: 137-145.
[15] Rempała G A, esołowski J. Double Asymptotics for the Chi-square Statistic[J]. Statistics & Probability Letters (S0167-7152), 2016, 119: 317-325.
[16] Ikram S T, Cherukuri A K.Intrusion Detection Model Using Fusion of Chi-square Feature Selection and Multi Class SVM[J]. Journal of King Saud University - Computer and Information Science (S1319-1578), 2017, 29(4): 462-472.
[17] Jadhav S, He H M, Jenkins K.Information Gain Directed Genetic Algorithm Wrapper Feature Selection For Credit Rating[J]. Applied Soft Computing (S1568-4946), 2018, 69: 541-553.
[18] Bennasar M, Hicks Y, Setchi R.Feature Selection Using Joint Mutual Information Maximisation[J]. Expert Systems with Applications (S0957-4174), 2015, 42(22): 8520-8532.
[19] 王强, 关毅, 王晓龙. 基于特征类别属性分析的文本分类器分类噪声裁减算法[J]. 自动化学报, 2007, 33(8): 809-816.
Wang Qiang, Guan Yi, Wang Xiaolong.A Method for Eliminating Class Noise In Text Classification Based on Feature Class Attribute[J]. Acta Automatica Sinica, 2007, 33(8): 809-816.

基于相对贡献率的噪声裁剪算法

Noise Clipping Algorithm Based on Relative Contribution Rate

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 2

编辑推荐

Metrics

本文评价

[1]	李晟, 李玉晓. 基于特征选择验证的故障注入仿真可信度评估[J]. 系统仿真学报, 2020, 32(8): 1567-1576.
[2]	谷雨, 张琴, 徐英. 基于多波段多极化仿真图像的SAR目标识别研究[J]. 系统仿真学报, 2017, 29(10): 2482-2488.