不完备信息系统中的多重代价决策粗糙集

南京大学学报(自然科学版) ›› 2015, Vol. 51 ›› Issue (2): 335–342.

不完备信息系统中的多重代价决策粗糙集

马兴斌1，鞠恒荣1,2,3，杨习贝1,2,3，宋晶晶1,2,3

出版日期:2015-03-06 发布日期:2015-03-06
作者简介:1. 江苏科技大学计算机科学与工程学院镇江212003; 2. 人工智能四川省重点实验室自贡643000; 3. 高维信息智能感知与系统教育部重点实验室南京210094)
基金资助:
国家自然科学基金(61100116,61272419), 江苏省自然科学基金 (BK2011492,BK2012700,BK20130471) , 高维信息智
能感知与系统教育部重点实验室( 南京理工大学) 开放基金(30920130122005) , 人工智能四川省重点实验室开放基金(13KJB520003,13KJD520008 )

Multi-cost based decision-theoretic rough sets in incomplete information systems

Ma Xingbin1, Ju Hengrong1,2,3, Yang Xibei1,2,3*, Song Jingjing1,2,3

Online:2015-03-06 Published:2015-03-06
About author:(1. School of Computer Science and Engineering, Jiangsu University of Science and Technology, Zhenjiang, 212003, China; 2. Artificial Intelligence Key Laboratory of Sichuan Province, Zigong, 643000, China; 3. Key Laboratory of Intelligent Perception and Systems for High-Dimensional Information (Nanjing University of Science and Technology), Ministry of Education, Nanjing, 210094, China)

摘要/Abstract

摘要： 决策粗糙集源于贝叶斯决策准则, 利用代价矩阵生成了构建概率粗糙集所需的一对阈值 . 通过代价对目标事物的近似使得决策粗糙集模型对代价敏感, 使决策粗糙集成为一种十分重要的粗糙集方法 . 然而, 大多数的决策粗糙
集模型仅使用一个代价矩阵进行描述, 来求解完备信息系统中的问题 . 这种方法并未考虑机器学习和数据挖掘中的一个重要问题, 即现实生活中代价本身所具有的多样与变化特性 . 为解决该问题, 首先, 通过使用多重代价矩阵将多
代价策略引进决策粗糙集; 然后, 在不完备信息系统中, 分别提出了乐观与悲观两种形式的多代价决策粗糙集方法, 讨论了这两种新的决策粗糙集模型与基于单代价矩阵决策粗糙集模型之间的关系, 并且给出了乐观和悲观这两种
决策粗糙集决策代价的总代价计算公式 . 最后, 在四组 U C I 数据集上对几种不同的决策粗糙集的决策代价进行了对比分析 . 实验结果表明, 乐观决策粗糙集得到的决策代价是一种较优的代价, 并且随着代价矩阵的增加, 代价的
值将会保持在一个稳定的值 . 揭示了决策理论粗糙集的潜在应用并且为其提供了新的研究方向 .

Abstract: Decision-theoretic rough set comes from Bayesian decision procedure, in which a pair of the thresholds is derived by the cost matrix for the construction of probabilistic rough set. Decision-theoretic rough set is a crucial rough set approach. By introducing the cost into probabilistic approximation of the target, the model of decision-theoretic rough set is actually sensitive to cost. However, most of the previous results about decision-theoretic rough set only use one and only one cost matrix to deal with the problems of the complete information systems. This method does not take the property of multiplicity and variability of cost into consideration, which is an important issue in machine learning and data mining. To solve such problems, a multi-cost strategy is firstly introduced into decision-theoretic rough set by using multiple cost matrixes. Moreover, the optimistic and pessimistic multi-cost decision-theoretic rough set models are proposed in incomplete information systems, respectively. Furthermore, the relationships are discussed between the two new decision-theoretic rough sets and the single cost matrix based decision-theoretic rough set. Then, we describe the formulas of the whole decision costs of optimistic and pessimistic multi-cost decision-theoretic rough set models. Finally, the several different decision costs of multi-cost decision-theoretic rough sets determined by decision-theoretic rough sets are tested on four UCI data sets. Experimental results show that the optimistic multi-cost decision-theoretic rough set model can generate the lowest decision cost. With the increase of the cost matrixes, all kinds of the whole decision costs will keep a steady value at last. The study suggests potential application areas and new research trends concerning decision-theoretic rough set.

马兴斌1，鞠恒荣1,2,3，杨习贝1,2,3，宋晶晶1,2,3. 不完备信息系统中的多重代价决策粗糙集[J]. 南京大学学报(自然科学版), 2015, 51(2): 335–342.

Ma Xingbin1, Ju Hengrong1,2,3, Yang Xibei1,2,3*, Song Jingjing1,2,3. Multi-cost based decision-theoretic rough sets in incomplete information systems[J]. Journal of Nanjing University(Natural Sciences), 2015, 51(2): 335–342.

参考文献

[1] Pawlak Z. Rough Sets- theoretical aspects of reasoning about data. Dordrecht: Kluwer Academic, 1991: 229.
[2] Hu Q H, Che X J, Zhang L, et al. Rank entropy based decision trees for monotonic classification. IEEE Transactions on Knowledge and Data Engineering, 2012, 24(11): 2052~2064.
[3] Hu Q H, Pan W W, Zhang L, et al. Feature selection for monotonic classification. IEEE Transactions on Fuzzy Systems, 2012, 20(1): 69~81.
[4] Luo G Z, Yang X B. Limited dominance-based rough set model and knowledge reductions in incomplete decision system. Journal of Information Science and Engineering, 2010, 26(6): 2199~2211.
[5] Yang X B, Yang J Y. Incomplete information system and rough set theory: Models and attribute reductions. Beijing: Science Press & Springer, 2012: 232.
[6] Kryszkiewicz M. Rough set approach to incomplete information systems. Information Sciences, 1998, 112(1-4): 39~49.
[7] 王国胤. Rough集理论在不完备信息系统中的扩充. 计算机研究与发展, 2002, 39(10): 1238~1243.
[8] 杨习贝, 杨静宇, 於东军等. 不完备信息系统中的可变精度分类粗糙集模型. 系统工程理论与实践, 2008, 28(5): 116~121.
[9] Yang X B, Qi Y S, Song X N, et al. Test cost sensitive multigranulation rough set: Model and minimal cost selection. Information Sciences, 2013, 250: 184~199.
[10] Min F, He H P, Qian Y H, et al. Test-cost-sensitive attribute reduction. Information Sciences, 2011, 181(22): 4928~4942.
[11] Min F, Liu Q H. A hierarchical model for test-cost-sensitive decision systems. Information Sciences, 2009, 179(14): 2442~2452.
[12] Min F, Zhu W. Attribute reduction of data with error ranges and test costs. Information Sciences, 2012, 211: 48~67.
[13] Min F, Hu Q H, Zhu W. Feature selection with test cost constraint. International Journal of Approximate Reasoning, 2014, 55(1): 167~179.
[14] Yao Y Y, Wong S K M, Lingras P. A decision-theoretic rough set model. In: The 5th International Symposium on Methodologies for Intelligent Systems, North-Holland, New York, 1990: 17~25.
[15] Liu D, Li T R, Li H X. A multiple-category classification approach with decision-theoretic rough sets. Fundamenta Informaticae, 2012, 115(2-3): 173~188.
[16] 刘盾, 李天瑞, 李华雄. 粗糙集理论: 基于三支决策视角. 南京大学学报（自然科学）, 2013, 49(5): 574~581.
[17] 谢聘, 商琳. 基于三支决策粗糙集的视频异常行为检测. 南京大学学报（自然科学）, 2013, 49(4): 475~482.
[18] Jia X Y, Liao W H, Tang Z M, et al. Minimum cost attribute reduction in decision-theoretic rough set models. Information Sciences, 2013, 219: 151~167.
[19] Yao Y Y, Zhao Y. Attribute reduction in decision-theoretic rough set models. Information Sciences, 2008, 178: 3356~3373.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed