شماره ركورد كنفرانس :
4109
عنوان مقاله :
Bayesian optimal design with stick-breaking priors
پديدآورندگان :
Goudarzi M Department of Statistics, Razi University, Kermanshah, Iran , Jafari H Department of Statistics, Razi University, Kermanshah, Iran , Khazaei S Department of Statistics, Razi University, Kermanshah, Iran
تعداد صفحه :
8
كليدواژه :
Dirichlet process , Bayesian optimal design , Random functionals , Stopping rule , Stick , breaking prior
سال انتشار :
1396
عنوان كنفرانس :
يازدهمين سمينار ملي احتمال و فرآيندهاي تصادفي
زبان مدرك :
انگليسي
چكيده فارسي :
In nonlinear regression models, the Fisher information matrix depends on unknown parameters and it is difficult to find an optimal design. So far, several methods have been proposed to solve this problem, such as Bayesian optimal designs. Bayesian optimal designs require some preliminary knowledge about the unknown parameters which might not be available in all applications. In the present paper, we assume the prior distribution is unknown and consider the nonparametric Bayesian approaches. The Dirichlet process is a fundamental tool in studying nonparametric Bayesian inference. We use Sethuraman s stick-breaking representation for approximating a distribution of random functional of Dirichlet process priors to introduce a Bayesian optimal design criterion. We study the result by Monte Carlo method in an example
كشور :
ايران
لينک به اين مدرک :
بازگشت