首页 | 本学科首页   官方微博 | 高级检索  
     检索      

CONTINUOUS-TIME KIEFER-WOLFOWITZ ALGORITHM WITH RANDOMIZED DIFFERENCES
作者姓名:CHEN Hanfu    WANG Qian
作者单位:Institute Of Systems Science,Academia Silica,Beijing 100080,China
摘    要:1.IntroductionTheKiefer-Wolfowitz(KW)algorithmisusedtosequeotiallysearchtheextremumofafUnction,onthebasisofthenoisecontaminatedobservationsofthefunction.AssumetheunknownauctionH:ac-Rhasauniquemaximumatxo.ItiswellknownthattheclassicalKWalgorithmneedsZmobserVationsofH(see1])forupdatingtheestimateofxoateachiteration.Itisrathertime-consumingifthedimensionmisverylarge,somanyresearchersmanagedtoreducethenumberofmeasurementsneededateachstep.Recently,Spalll2]proposedanalgorithmwithrandomizeddif…


CONTINUOUS-TIME KIEFER-WOLFOWITZ ALGORITHM WITH RANDOMIZED DIFFERENCES
CHEN Hanfu, WANG Qian.CONTINUOUS-TIME KIEFER-WOLFOWITZ ALGORITHM WITH RANDOMIZED DIFFERENCES[J].Journal of Systems Science and Complexity,1998(4).
Authors:CHEN Hanfu    WANG Qian
Abstract:A continuous-time Kiefer-Wolfowitz algorithm with randomized differences andwith truncations at randomly varying bounds is proposed. It is shown that the algorithmconverges to the desired value almost surely under mild conditions. The rate of convergenceand the asymptotic normality of the algorithm are also established.
Keywords:Continuous-time  stochastic approximation  random truncations  randomized differences  stochastic delay differential equation  asymptotic normality
本文献已被 CNKI 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号