广西师范大学学报（自然科学版） ›› 2011, Vol. 29 ›› Issue (2): 110-113.

• • 上一篇    下一篇

误差在Cost-Sensitive分类中的应用

廖元秀, 周生明   

  1. 广西师范大学计算机科学与信息工程学院,广西桂林541004
  • 收稿日期:2011-02-28 发布日期:2018-11-19
  • 通讯作者: 廖元秀(1963—),女,广西灵川人,广西师范大学副教授。E-mail:liaoyuanxiu@mailbox.gxnu.edu.cn
  • 基金资助:
    国家自然科学基金资助项目(60963008)

Application of Errors in Cost-Sensitive Classifications

LIAO Yuan-xiu, ZHOU Sheng-ming   

  1. College of Computer Science and Information Engineering,Guangxi Normal University,Guilin Guangxi 541004,China
  • Received:2011-02-28 Published:2018-11-19

摘要： 针对使检查代价和误分类代价最小化的Cost-Sensitive学习,讨论误差在分类过程中的应用,提出一个带阈值的决策树,并给出一个带阈值的检查策略。在基于Cost-Sensitive学习的分类中,确定属性值所用到的检测手段和设备精度存在一定的误差值,评估误分类代价更是有较大的误差。另外,很多分类问题并不要求达到百分之百的正确率,允许有一定的误差范围。把这些误差的边界看作是一个阈值,利用这种阈值来简化决策树的建立,改进检查策略的设计,提高分类效率。

关键词: Cost-Sensitive学习, 分类代价, 检查策略, 分类误差范围, 阈值

Abstract: Aiming at the minimization problem of test costs andmisclassification costs on Cost-Sensitive learning,application of errors in classifications are discussed.A kind of decision trees and test strategies with thresholdsare proposed.There are errors on both Cost-Sensitive classifications resaltedfrom methods of tests and equipment accuracy and evaluating misclassification.In addition,many classification problems are not required to achieve one hundred percent classification accuracy.The boundaries of these errors are regarded as a kind of threshold values.The establishment of decision trees is simplified and the design of test strategies and the classification efficiency are improved by using these threshold values.

Key words: Cost-Sensitive learning, misclassification costs, test strategies, error margin of classifications, thresholds

中图分类号: 

  • TP18
[1] NUNEZ M.The use of background knowledge in decision tree induction[J].Machine Learning,1991,6(3):231-250.
[2] TAN Ming.Cost-sensitive learning of classification knowledge andits applications in robotics[J].Machine Learning,1993,13(1):7-33.
[3] TURNEY P.Types of cost in inductive concept learning[C]//Workshop on Cost-Sensitive Learning at the Seven-teenth International Conference onMachine Learning.Stanford,CA:Stanford University,2000:15-21.
[4] ZUBEK V B,DIETTERICH T G.Pruning improves heuristic search for cost-sensitive learning[C]//Proceeding of the Nineteenth international Conference onMachine Learning.Sydney:Morgan Kaufmann Press,2002:27-34.
[5] GREINER R,GROVE A J,ROTH D.Learning cost-sensitive active classifiers[J].Artificial Intelligence,2002,139(2):137-174.
[6] LING C X,YANG Qiang,WANG Jian-ning,et al.Decision trees withminimal costs[C]//Proceedings of the Twenty-First International Conference onMachine Learning.New York:ACM Press,2004:69.
[7] CHAI Xiao-yong,DENG Lin,YANG Qiang,et al.Test-cost sensitivenaive bayesian classification[C]//Proceedings of the Fourth IEEE International Conference on Data Mining.Brighton:IEEE Computer Society Press,2004:51-58.
[8] 周生明,廖元秀.Cost-Sensitive学习的一个新课题[J].广西师范大学学报:自然科学版,2007,25(4):55-58.
[9] ZHANG Shi-chao,QIN Zhen-xing,LING C X,et al.Missing is useful:missing values in cost-sensitive decision trees[J].IEEE Transactions on Knowledge and Data Engineering,2005,17(12):1689-1693.
[1] 连天培, 蒋品群, 宋树祥, 蔡超波, 庞中秋. 低温度系数高电源抑制比宽频带带隙基准电压源的设计[J]. 广西师范大学学报（自然科学版）, 2019, 37(1): 125-132.
[2] 周克良, 邢素林, 聂丛楠. 基于自适应阈值小波变换的心音去噪方法[J]. 广西师范大学学报（自然科学版）, 2016, 34(1): 19-25.
[3] 华梓铮, 华泽玺. 基于NSCT的含噪图像边缘检测算法[J]. 广西师范大学学报（自然科学版）, 2014, 32(2): 26-34.
[4] 郝丽杰, 蒋贵荣, 鹿鹏. 具垂直传染的SIRS传染病模型的脉冲控制和分岔分析[J]. 广西师范大学学报（自然科学版）, 2012, 30(4): 42-47.
[5] 张新明, 张玉珊, 李振云. 一种改进的矩不变图像分割方法[J]. 广西师范大学学报（自然科学版）, 2011, 29(2): 185-190.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
版权所有 © 广西师范大学学报（自然科学版）编辑部
地址：广西桂林市三里店育才路15号　邮编：541004
电话：0773-5857325　E-mail: gxsdzkb@mailbox.gxnu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发