摘要
在ε-支持向量回归和粗糙v-支持向量回归模型的基础上,研究了新的粗糙ε-支持向量回归模型.利用固定对称边界粗糙ε-不敏感损失函数,得到粗糙ε-不敏感管,构造固定对称边界粗糙ε-支持向量回归模型;利用固定非对称边界粗糙-不敏感损失函数,得到粗糙εu-εd-不敏感管,构造固定非对称边界粗糙ε-支持向量回归模型.通过引进Lagrange函数和根据KKT条件,处理粗糙ε-支持向量回归模型的对偶问题.
Two new rough e-support vector regression models are proposed based on ε-support vector regression, rough v-support vector regression and rough set theory. Firstly,a rough boundary ε-insensitive tube is defined with fixed symmetrical boundary rough ε-insensitive loss function, the method of optimization and ε-support vector regression model. Moreover we design a fixed symmetrical boundary rough ε-support vector regression model(RFSM SVR). Secondly,we extend the model to the case that asymmetrical loss function is considered. Finally, according to Karush Kuhn Tucker(KKT)conditions, we derive their dual problems by introducing the Lagrange functional into rough ; support vector regression models.
出处
《南京大学学报(自然科学版)》
CAS
CSCD
北大核心
2013年第5期650-654,共5页
Journal of Nanjing University(Natural Science)
基金
国家自然科学基金(61170107
61222210)
河北省高等学校科学研究计划(Z2010188)