摘要
多粒度粗糙集模型是一种有效的信息融合策略。利用该策略能从多个角度将多源信息进行融合,并转化成一致的信息表示。现有的大多数多粒度信息融合方法对每个知识粒度都采用相同的阈值,然而,众所周知,不同的信息源的来源和噪声都不尽相同,其对应的知识粒度的阈值也应不同。为此,首先在广义多粒度粗糙集中引入单参数决策理论粗糙集,提出了广义自适应多粒度粗糙集模型。然后,利用经典的融合策略设计了4种广义多粒度模型,所有模型都可以通过一个参数补偿系数ζ来自适应地获得知识粒度对应的阈值对,并讨论了这些模型的相关性质。最后,通过实验结果证明,所提模型在实际应用中灵活性更高,决策更为合理。
The multi-granulation rough set model is an effective information fusion strategy. In this paper, this strategy is used to fuse multi-source information from multiple perspectives, and then this information is translated into a consistent information representation. However, most existing multi-granulation information fusion methods use the same threshold value for each knowledge granularity. As we all know, the origin and noise differ among information sources,and the threshold values of the corresponding knowledge granularity should differ. To this end, in this paper, a generalized adaptive multi-granulation rough set model is proposed by combining a single-parameter decision-theoretic rough set with a generalized multi-granulation rough set. Then, four types of generalized multi-granulation models are designed based on typical fusion strategies so that all models can obtain threshold pairs corresponding to knowledge granularity by setting a compensation coefficient ζ. Furthermore, the relevant properties of these models are discussed. Finally, the experimental results demonstrate that the proposed model is more flexible and reasonable in practical applications.
作者
钱进
童志钢
余鹰
洪承鑫
苗夺谦
QIAN Jin;TONG Zhigang;YU Ying;HONG Chengxin;MIAO Duoqian(College of Software Engineering,East China Jiaotong University,Nanchang 330013,China;College of Electronic and Information Engineering,Tongji University,Shanghai 201804,China)
出处
《智能系统学报》
CSCD
北大核心
2023年第1期173-185,共13页
CAAI Transactions on Intelligent Systems
基金
国家自然科学基金项目(62066014,62163016)
江西省自然科学基金项目(20202BABL202018,20212ACB202001)。
关键词
多源信息
信息融合
决策理论粗糙集
粗糙集
广义自适应多粒度
多粒度
自适应阈值
知识粒度
multi-source information
information fusion
decision-theoretic rough set
rough set
generalized adaptive multi-granulation
multi-granulation
adaptive thresholds
knowledge granularity
