摘要
模糊认知图(fuzzy cognitive map,FCM)具有简单的推理机制和较强的因果关系表达能力,已得到广泛关注和研究,但FCM对专家经验知识具有较强的依赖性,故而限制了在复杂动态系统建模中的应用。基于此,提出了一种测度递进策略的模糊认知图学习方法。利用线性回归算法,学习得到模糊认知图权重矩阵粗模型;将神经网络的权值调整算法应用于权重矩阵粗模型的细化过程,将该模糊认知图模型应用在股票市场,实现对股票日均值的预测。实验结果表明了该建模方式是有效的。
Fuzzy cognitive map has been received extensive attentions due to its simple reasoning mechanism and strong causal relationship expression,but its reliance on expert experience and knowledge limits its application in the complex dynamic system.Based on this,a progressive strategy fuzzy cognitive maps learning method is proposed.First,linear regression algorithm is used to obtain the coarse model of weight matrix of fuzzy cognitive maps;then the neural network weight adjustment algorithm is applied to the refinement process of the coarse weight matrix model.Finally,this model is used in forecasting the daily average price in the stock market.Experiments show that this modeling approach is effective.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第5期1958-1962,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(61175048)
北京市属高等学校人才强教计划资助基金项目
关键词
模糊认知图
推理机制
测度递进
股票
预测
FCM
inference mechanism
progressive measure
stock
prediction