一种基于最大最小独立性的因果发现算法
发布时间：2018-11-02 19:12
【摘要】：线性非高斯无环模型(LiNGAM)具有在没有任何先验知识的情况下能够仅仅从观察数据中完整地识别因果网络的优势,这使得它得到了越来越多研究者的关注.然而,现有求解LiNGAM模型的算法中一部分存在对初始值敏感,容易陷入局部最优解的问题,一部分存在对于外生变量识别率低的缺陷.为此,提出了一种基于最大最小独立性的因果发现算法.通过引入自适应的独立性判定参数,根据此参数来找出与其余所有变量回归得到的残差都独立的变量,即为外生变量.该算法不仅避免了传统算法对独立性值差异敏感而导致识别率低的问题,而且也避免了不同数据集对固定独立性参数敏感而导致无法识别的缺陷.将该算法应用于虚拟网络和真实网络中,实验结果都表明,各种维度下该算法都优于现有的其他算法.
[Abstract]:The linear non-Gao Si acyclic model (LiNGAM) has the advantage of only recognizing causal networks completely from observational data without any prior knowledge, which has attracted more and more researchers' attention. However, some of the existing algorithms for solving the LiNGAM model are sensitive to the initial value and easily fall into the local optimal solution, while others have the defect of low recognition rate for exogenous variables. Therefore, a causal discovery algorithm based on maximum and minimum independence is proposed. By introducing an adaptive independence decision parameter, the variables which are independent of the residuals obtained from the regression of all the other variables are found according to this parameter, that is, the exogenous variables. This algorithm not only avoids the problem of low recognition rate caused by the sensitivity of traditional algorithms to the difference of independence value, but also avoids the defect that different data sets are sensitive to fixed independence parameters and can not be recognized. The experimental results show that the algorithm is superior to other existing algorithms in different dimensions.
【作者单位】： 广东工业大学;佛山科学技术学院;
【基金】：NSFC-广东联合基金(U1501254) 国家自然科学基金(61472089,61572143) 广东省自然科学基金(2014A030306004,2014A030308008) 广东省科技计划项目(2013B051000076,2015B010108006,2015B010131015) 广东特支计划(2015TQ01X140) 广州市珠江科技新星(201610010101)
【分类号】：TP18
本文编号：2306704
[Abstract]:The linear non-Gao Si acyclic model (LiNGAM) has the advantage of only recognizing causal networks completely from observational data without any prior knowledge, which has attracted more and more researchers' attention. However, some of the existing algorithms for solving the LiNGAM model are sensitive to the initial value and easily fall into the local optimal solution, while others have the defect of low recognition rate for exogenous variables. Therefore, a causal discovery algorithm based on maximum and minimum independence is proposed. By introducing an adaptive independence decision parameter, the variables which are independent of the residuals obtained from the regression of all the other variables are found according to this parameter, that is, the exogenous variables. This algorithm not only avoids the problem of low recognition rate caused by the sensitivity of traditional algorithms to the difference of independence value, but also avoids the defect that different data sets are sensitive to fixed independence parameters and can not be recognized. The experimental results show that the algorithm is superior to other existing algorithms in different dimensions.
【作者单位】： 广东工业大学;佛山科学技术学院;
【基金】：NSFC-广东联合基金(U1501254) 国家自然科学基金(61472089,61572143) 广东省自然科学基金(2014A030306004,2014A030308008) 广东省科技计划项目(2013B051000076,2015B010108006,2015B010131015) 广东特支计划(2015TQ01X140) 广州市珠江科技新星(201610010101)
【分类号】：TP18
【相似文献】
相关期刊论文 前1条
1 刘佳昆;周永权;;一种最大最小萤光素值人工萤火虫算法[J];计算机应用研究;2011年10期
相关硕士学位论文 前3条
1 何凝;基于最大最小距离法的音乐节拍跟踪算法研究[D];天津大学;2014年
2 马文华;改进最大最小蚁群算法的研究及应用[D];安徽大学;2015年
3 张晓丹;WSN中基于改进粒子群优化算法的分簇拓扑算法研究[D];郑州大学;2017年
，本文编号：2306704
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2306704.html
