多尺度决策系统中代价敏感的最优尺度组合  

Cost-sensitive optimal scale combination in multi-scale decision systems

作　　者：张清华[1,2] 张雪秋 庞国弘 ZHANG Qing-hua;ZHANG Xue-qiu;PANG Guo-hong(School of Computer Science and Technology,Chongqing University of Posts and Telecommunications,Chongqing 400065,China;Chongqing Key Laboratory of Computational Intelligence,Chongqing University of Posts and Telecommunications,Chongqing 400065,China)

机构地区：[1]重庆邮电大学计算机科学与技术学院,重庆400065 [2]重庆邮电大学计算智能重庆市重点实验室,重庆400065

出　　处：《控制与决策》2021年第10期2369-2378,共10页Control and Decision

基　　金：国家自然科学基金项目(61876201)。

摘　　要：最优尺度组合是多尺度决策系统中的研究热点之一,现有的研究大多是从一致性、不确定性的角度出发,而没有充分考虑代价信息的影响.针对该问题,首先分析最优尺度组合中考虑代价敏感的重要性,从决策代价的角度提出基于测试代价和延迟代价的多尺度决策系统,并且定义了尺度代价和属性代价来刻画尺度和属性所产生的代价;其次,考虑实际场景中属性代价的影响,将属性重要度和属性代价结合进行属性排序;最后,在属性进行最优尺度选择时,考虑尺度代价的影响,建立了一个代价敏感的最优尺度组合选择模型.实验结果表明,在现有代价认知场景下,所提模型能合理地进行最优尺度组合选择,所得结果更符合实际需求.Optimal scale combination is one of the research hotspots in multi-scale decision systems. Most existing researches start from the perspective of consistency and uncertainty, without fully considering the influence of cost information. In view of this issue, the importance of considering cost-sensitivity in optimal scale combination is analyzed. Firstly, a multi-scale decision system based on test cost and delay cost from the perspective of decision cost is proposed, and the scale cost and attribute cost are defined to describe the cost caused by scale and attribute. Then,considering the effect of attribute cost in the real scene, attribute significance and attribute cost are combined to sort the attributes. Finally, a cost-sensitive optimal scale combination selection model is established by considering the influence of scale cost when selecting the optimal scale of attributes. The experimental results show that the proposed model can reasonably select the optimal scale combination with the existing cost cognition scenario, which is more in line with the actual requirements.

关 键 词：多尺度决策系统 属性重要度 代价敏感 最优尺度组合 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]