Title :
Single-run gradient estimation via measure-valued differentiation
Author :
Heidergott, Bernd ; Hordijk, Arie
Author_Institution :
Tinbergen Inst., Vrije Univ. Amsterdam, Netherlands
Abstract :
We show how single-run-based measure-valued differentiation gradient estimators can be obtained. The key idea is to apply a change of measure a posterior to the mathematical analysis of the derivative. From the point of view of the likelihood ratio method, we show that likelihood ratio type gradient estimators can be applied in situations where the mathematical conditions needed for applying a likelihood ratio analysis are not met.
Keywords :
Markov processes; differentiation; gradient methods; maximum likelihood estimation; Markov chains; likelihood ratio method; measure-valued differentiation; single-run gradient estimation; Density measurement; Imaging phantoms; Kernel; Mathematical analysis; State estimation; Stochastic systems; Sufficient conditions; Telecommunication computing; Transportation; Virtual manufacturing; Gradient estimation; Markov chains; likelihood ratios; measure-valued differentiation;
Journal_Title :
Automatic Control, IEEE Transactions on
DOI :
10.1109/TAC.2004.835588
