求解线性模型稳健参数估计的整数编码遗传算法

线性模型回归系数的一些稳健估计如LMS、LQS、LTS、LTA的应用越来越广泛,然而它们的精确计算依赖于NP难题,在遇到高维大规模数据集时不可能在较短时间内得到精确解.为尽快得到较高精度的近似解,提出了求解线性模型的稳健参数估计的整数编码遗传算法,通过计算机模拟试验验证了算法可以更快地找出全局最优解.

Integer-coded Genetic Algorithm for Robust Parameter Estimation of Linear Regression Model

The linear regression coefficient of some robust estimation such as LMS,LTS, LTA has been widely applied,but their exact calculation depends on the NP problem.The exact solutions for large high dimensional data set can't compute in a reasonable time.To get more accurate results as soon as possible,the approximate solution using integer-coded genetic algorithm for solving linear model estimation is introduced.The computer simulation shows that the algorithm is correct and effective.