Convergence of parameter sensitivity estimates in a stochastic experiment

From MaRDI portal





There is a serious problem in stochastic optimization theory. It is the problem to estimate the gradient (parameter sensitivity) of an unknown function. The author deals exactly with this problem. More precisely, he studies the estimation possibilities of the function gradient I(\(\vartheta)\) given by \(I(\vartheta)=E[L(\vartheta,\xi)]\) where L(\(\cdot,\cdot)\) is a real valued function, \(\vartheta \in R^ n\), \(\xi\) is a random vector and E denotes the operator of expectation. Three types of gradient estimation are presented in the paper. The greatest attention is given to the method using the strong law of large numbers and the possibility to exchange the expectation operator and the differentiation operator.




Cited in
(28)








This page was built for publication: Convergence of parameter sensitivity estimates in a stochastic experiment

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3692660)