南京大学学报(自然科学版) ›› 2023, Vol. 59 ›› Issue (6): 981–995.doi: 10.13232/j.cnki.jnju.2023.06.008

• • 上一篇    下一篇

分层多尺度决策信息系统的序贯三支决策

刘芳1, 李磊军1,2(), 米据生1,2, 李美争3   

  1. 1.河北师范大学数学科学学院,石家庄,050024
    2.河北省计算数学与应用重点实验室,石家庄,050024
    3.河北师范大学计算机与网络空间安全学院,石家庄,050024
  • 收稿日期:2023-08-10 出版日期:2023-11-30 发布日期:2023-12-06
  • 通讯作者: 李磊军 E-mail:lileijun1985@163.com
  • 基金资助:
    国家自然科学基金(61502144);河北省自然科学基金(F2018205196);河北省高等学校科学技术研究项目(BJ2019014)

Sequential three⁃way decision of hierarchical multi⁃scale decision information system

Fang Liu1, Leijun Li1,2(), Jusheng Mi1,2, Meizheng Li3   

  1. 1.School of Mathematical Sciences, Hebei Normal University, Shijiazhuang, 050024, China
    2.Hebei Key Laboratory of Computational Mathematics and Applications, Shijiazhuang, 050024, China
    3.School of Computer and Cyberspace Security, Hebei Normal University, Shijiazhuang, 050024, China
  • Received:2023-08-10 Online:2023-11-30 Published:2023-12-06
  • Contact: Leijun Li E-mail:lileijun1985@163.com

摘要:

传统信息系统中每个属性只有一个尺度,但随着海量数据的涌现,实际应用中经常是在多个尺度上处理和分析问题.三支决策是解决分类问题的一种经典方法,序贯三支决策是在三支决策的基础上进行多步决策的一种方法.将多尺度决策信息系统与三支决策相结合,基于决策理论粗糙集提出分层多尺度决策信息系统的序贯三支决策模型,得到动态变化的正域、负域、边界域.对多尺度决策信息系统进行分层,依次在分层后得到的多个单尺度决策信息系统上进行讨论,构建尺度层面的序贯结构;在每个单尺度决策信息系统上,通过增加属性的方式得到属性子集序列,诱导出多级粒度结构,构建该尺度下粒度层面的序贯结构.为此,给出两种属性子集序列的选择方法;在序贯三支决策过程中,利用相对损失函数计算阈值,并讨论了阈值的性质;最后给出序贯三支决策过程中的分类规则,并用实例说明提出的模型能有效地处理分类问题.

关键词: 多尺度信息系统, 序贯三支决策, 粗糙集, 分层, 分类

Abstract:

In traditional information systems,each attribute has only one scale,but with the emergence of massive data,we often deal with and analyze problems on multiple scales in practical applications. Three?way decision (3WD) is a classical method to solve the classification problem. Sequential three?way decision (S3WD) is a method of multi?step decision?making based on 3WD. This paper combines multi?scale decision information system with 3WD. Based on decision?theoretic rough sets (DTRSs),the sequential three?way decision model of hierarchical multi?scale decision information system is proposed. After that,the dynamic positive region,negative region and boundary region can be obtained. The multi?scale decision information system is layered. It is discussed in turn on multiple single?scale decision information systems obtained after stratification,and the sequential structure of the scale level is constructed. On each single?scale decision information system,the attribute subset sequence is obtained by adding attributes,and the multilevel granular structure is induced to construct the sequential structure at the granularity level. For this reason,this paper gives two selection methods for the collection of attribute subsets,and in the S3WD process,the relative loss function is used to calculate the threshold,and the properties of the threshold are discussed. Finally,the classification rules in the S3WD process are given,and an example is given to illustrate that the proposed model can effectively deal with the classification problem.

Key words: multi?scale information system, sequential three?way decision, rough set, hierarchical, classification

中图分类号: 

  • TP18

表1

损失函数"

DP¬DN
aPλPPλPN
aBλBPλBN
aNλNPλNN

表2

相对损失函数"

DP¬DN
aPλPP'=0λPN'=λPN-λNN
aBλBP'=λBP-λPPλBN'=λBN-λNN
aNλNP'=λNP-λPPλNN'=0

表3

由属性值导出的相对损失函数"

xiaj¬aj
aP0maxj-xij
aBσxij-minjσmaxj-xij
aNxij-minj0

图1

多尺度决策信息系统S上的序贯三支决策过程"

表4

信息系统表"

Ua1a2amd
x1x11x12x1md1
x2x21x22x2md2
xnxn1xn2xnmdn

表5

用等价类刻画距离"

Ua1a2am
x1x11x12x1m
x2x21x22x2m
xnxn1xn2xnm

表6

属性之间的距离"

a1a2am
a1d11d12d1m
a2d21d22d2m
amdm1dm2dmm

表7

信息系统S"

Ua1a2a3a4a5a6d
x11011011
x20011010
x31101010
x41101011
x50000100
x61011000
x71000111
x81110001

表8

信息系统S中属性之间的距离"

a1a2a3a4a5a6
a100.470.50.470.3750.47
a20.4700.470.50.470.5
a30.50.4700.470.3750.47
a40.470.50.4700.220.375
a50.3750.470.3750.2200.47
a60.470.50.470.3750.470

表9

聚类结果"

属性a1a2a3a4a5a6
类别G4G4G1G3G3G2

图2

单尺度决策信息系统Sh上的序贯三支决策过程"

图3

序贯三支决策过程中状态集合的变化"

表10

基于属性值的相对损失函数"

λijhcjh¬cjh
aPλij,PPh=0λij,PNh=maxjh-xijh
aBλij,BPh=σjhxijh-minjhλij,BNh=σjhmaxjh-xijh
aNλij,NPh=xijh-minjhλij,NNh=0

表11

相对损失函数的汇总"

c1hc2hcqh
x1λ11hλ12hλ1qh
x2λ21hλ22hλ2qh
xnλn1hλn2hλnqh

表12

多尺度决策信息系统S"

Ua1a2a3a4d
a11a12a21a22a31a32a41a42
x1456532650
x2556578851
x3787878651
x4454555651
x5555578451
x6655565781
x7222245320
x8788865550
x9653265650
x10223245550

表13

单尺度决策信息系统S1"

Ua11a21a31a41d
x146360
x256781
x377761
x444561
x555741
x665671
x722430
x878650
x963660
x1023450

表14

单尺度决策信息系统S2"

Ua12a22a32a42d
x155250
x255851
x388851
x455551
x555851
x655581
x722520
x888550
x952550
x1022550

表15

S2中属性值的最大值、最小值及σj2的值"

a12a32a42
maxj2888
minj2222
σj20.20.2250.25

表16

A12粒度下的相对损失函数"

D12¬D12
λPPλBPλNPλPNλBNλNN
x100.6330.60
x301.26000
x700061.20

表17

A12粒度下的阈值"

αβγ
x10.80.20.5
x3000
x7111

表18

S1中属性值的最大值、最小值及σj1的值"

a11a21a41
maxj1788
minj1223
σj10.30.3250.35
1 Pawlak Z. Rough sets. International Journal of Computer & Information Sciences198211(5):341-356.
2 Yao Y Y. Probabilistic rough set approximations. International Journal of Approximate Reasoning200849(2):255-271.
3 Yao Y Y. Decision?theoretic rough set models∥The 2nd International Conference on Rough Sets and Knowledge Technology. Springer Berlin Heidelberg,2007:1-12.
4 Yao Y Y. Three?way decisions with probabilistic rough sets. Information Sciences2010180(3):341-353.
5 Huang X F, Zhan J M, Sun B Z. A three?way decision method with pre?order relations. Information Sciences2022(595):231-256.
6 Jia F, Liu P D. A novel three?way decision model under multiple?criteria environment. Information Sciences2019(471):29-51.
7 Zhang C Y, Feng X Z, Gao R Y. Three?way decision models and its optimization based on Dempster–Shafer evidence theory and rough sets. Granular Computing20216(2):411-420.
8 Yao Y Y. Granular computing and sequential three?way decisions∥The 8th International Conference on Rough Sets and Knowledge Technology. Springer Berlin Heidelberg,2013:16-27.
9 Xu Y, Tang J X, Wang X S. Three sequential multi?class three?way decision models. Information Sciences2020(537):62-90.
10 Qian J, Dang C Y, Yue X D,et al. Attribute reduction for sequential three?way decisions under dynamic granulation. International Journal of Approximate Reasoning2017(85):196-216.
11 Zhang Q H, Pang G H, Wang G Y. A novel sequential three?way decisions model based on penalty function. Knowledge?Based Systems2020(192):105350.
12 Li H X, Zhang L B, Zhou X Z,et al. Cost?sensitive sequential three?way decision modeling using a deep neural network. International Journal of Approximate Reasoning2017(85):68-78.
13 Qian J, Liu C H, Miao D Q,et al. Sequential three?way decisions via multi?granularity. Information Sciences2020(507):606-629.
14 Qian J, Tang D W, Yu Y,et al. Hierarchical sequential three?way decision model. International Journal of Approximate Reasoning2022(140):156-172.
15 Wu W Z, Leung Y. Theory and applications of granular labelled partitions in multi?scale decision tables. Information Sciences2011181(18):3878-3897.
16 Wu W Z, Leung Y. Optimal scale selection for multi?scale decision tables. International Journal of Approximate Reasoning201354(8):1107-1129.
17 Zheng J W, Wu W Z, Bao H,et al. Evidence theory based optimal scale selection for multi?scale ordered decision systems. International Journal of Machine Learning and Cybernetics202213(4):1115-1129.
18 Chen Y, Hu J, Zhang Q H,et al. Multi?scale set?valued information system and its optimal scale selection. Journal of Shanxi University (Natural Science Edition)202043(4):765-775.
19 Li F, Hu B Q. A new approach of optimal scale selection to multi?scale decision tables. Information Sciences2017(381):193-208.
20 张嘉茹,吴伟志,杨烨. 协调广义决策多尺度序信息系统的知识获取. 模式识别与人工智能202235(9):789-804.
Zhang J R, Wu W Z, Yang Y. Know?ledge acquisition for consistent generalized decision multi?scale ordered information systems. Pattern Recognition and Artificial Intelligence202235(9):789-804.
21 胡军,陈艳,张清华,等. 广义多尺度集值决策系统最优尺度选择. 计算机研究与发展202259(9):2027-2038.
Hu J, Chen Y, Zhang Q H,et al. Optimal scale selection for generalized multi?scale set?valued decision systems. Journal of Computer Research and Development202259(9):2027-2038.
22 Deng J, Zhan J M, Wu W Z. A three?way decision methodology to multi?attribute decision?making in multi?scale decision information systems. Information Sciences2021(568):175-198.
23 Hao C, Li J H, Fan M,et al. Optimal scale selection in dynamic multi?scale decision tables based on sequential three?way decisions. Information Sciences2017(415-416):213-232.
24 Luo C, Li T R, Huang Y Y,et al. Updating three?way decisions in incomplete multi?scale information systems. Information Sciences2019(476):274-289.
25 Cheng Y L, Zhang Q H, Wang G Y,et al. Optimal scale selection and attribute reduction in multi?scale decision tables based on three?way decision. Information Sciences2020(541):36-59.
26 Chen Y M, Qin N, Li W,et al. Granule structures,distances and measures in neighborhood systems. Knowledge?Based Systems2019(165):268-281.
27 陈德刚,徐伟华,李金海,等. 粒计算基础教程. 北京:科学出版社,2019,126.
28 Lu Z C, Qin Z, Zhang Y Q,et al. A fast feature selection approach based on rough set boundary regions. Pattern Recognition Letters2014(36):81-88.
[1] 金铭, 陈锦坤, 孙亚超. 基于边界域条件熵的最优尺度约简[J]. 南京大学学报(自然科学版), 2023, 59(6): 1034-1047.
[2] 程凤伟, 王文剑, 张珍珍. 面向高维小样本数据的层次子空间ReliefF特征选择算法[J]. 南京大学学报(自然科学版), 2023, 59(6): 928-936.
[3] 蒋浩英, 钱进, 王滔滔, 洪承鑫, 余鹰. 基于三支决策的新型分类匿名模型[J]. 南京大学学报(自然科学版), 2023, 59(6): 970-980.
[4] 张义宗, 王磊, 徐阳. 属性集变化下序决策信息系统的增量属性约简算法[J]. 南京大学学报(自然科学版), 2023, 59(5): 813-822.
[5] 韩雪, 周晨. 大气探测激光雷达的分类和特征[J]. 南京大学学报(自然科学版), 2023, 59(5): 900-913.
[6] 孟元, 张轶哲, 张功萱, 宋辉. 基于特征类内紧凑性的不平衡医学图像分类方法[J]. 南京大学学报(自然科学版), 2023, 59(4): 580-589.
[7] 王冰洁, 张超, 李德玉, 马瑾男, 王渊. 基于区间二型模糊多粒度证据融合方法的钢铁行业耗能决策[J]. 南京大学学报(自然科学版), 2023, 59(4): 600-609.
[8] 谭嘉辰, 董永权, 张国玺. SSM: 基于孪生网络的糖尿病视网膜眼底图像分类模型[J]. 南京大学学报(自然科学版), 2023, 59(3): 425-434.
[9] 吕佳, 肖锋. 内存有效的快速双层深度规则分类器[J]. 南京大学学报(自然科学版), 2023, 59(3): 446-459.
[10] 冯海, 马甲林, 许林杰, 杨宇, 谢乾. 融合标签嵌入和知识感知的多标签文本分类方法[J]. 南京大学学报(自然科学版), 2023, 59(2): 273-281.
[11] 徐伟华, 潘彦舟. 加权变精度直觉模糊序信息决策表的近似约简[J]. 南京大学学报(自然科学版), 2023, 59(1): 1-11.
[12] 于子淳, 吴伟志. 具有多尺度决策的信息系统的最优全局剪枝选择[J]. 南京大学学报(自然科学版), 2023, 59(1): 12-21.
[13] 陈瑞, 徐金东, 刘兆伟, 阎维青, 王璇, 宋永超, 倪梦莹. 基于模糊空谱特征的高光谱图像分类[J]. 南京大学学报(自然科学版), 2023, 59(1): 145-154.
[14] 田小瑜, 秦永彬, 黄瑞章, 陈艳平. 基于相关性约束矩阵分解的多标签分类方法[J]. 南京大学学报(自然科学版), 2023, 59(1): 76-84.
[15] 张艳莎, 冯夫健, 王杰, 潘凤, 谭棉, 张再军, 王林. 基于张量特征的小样本图像快速分类方法[J]. 南京大学学报(自然科学版), 2022, 58(6): 1059-1069.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!