On Finite Stages Stochastic Approximation with Some Empirical Studies

碩士 === 淡江大學 === 數學研究所 === 70 ===   Let M1(x) and M2(x) be respectively two unknown regbression functions (response function with levelx). Three kinds of problems are considered.   (1) For given α, let θ1, and θ2 be respectively the unique root of M1 and M2, I.e. M1(θ1)=M2(θ2)=α. By taking samples...

Full description

Bibliographic Details
Main Authors: Su, Yu-Wen, 蘇郁文
Other Authors: Huang, Wen-Tao
Format: Others
Language:zh-TW
Online Access:http://ndltd.ncl.edu.tw/handle/97377841164907777274
id ndltd-TW-070TKU03479005
record_format oai_dc
spelling ndltd-TW-070TKU034790052016-02-15T04:13:33Z http://ndltd.ncl.edu.tw/handle/97377841164907777274 On Finite Stages Stochastic Approximation with Some Empirical Studies 有限步驟的隨機逼近及其慕擬研究 Su, Yu-Wen 蘇郁文 碩士 淡江大學 數學研究所 70   Let M1(x) and M2(x) be respectively two unknown regbression functions (response function with levelx). Three kinds of problems are considered.   (1) For given α, let θ1, and θ2 be respectively the unique root of M1 and M2, I.e. M1(θ1)=M2(θ2)=α. By taking samples, we want to decideθ1≧θ2 or θ1<θ2.   (2) For given α, we propose an algorithm to find a confidence interval for θ, with a prefixed confidence probability, where M1(θ)=α.   (3) Assume a unique θ such that M1(θ) attains its maximum, we propose an algorithm to find a confidence interval for θ with a prefixed confidence probability.   By using computer, some Monte Carlo Studies are made for the proposed algorithm. Huang, Wen-Tao 黃文濤 學位論文 ; thesis 30 zh-TW
collection NDLTD
language zh-TW
format Others
sources NDLTD
description 碩士 === 淡江大學 === 數學研究所 === 70 ===   Let M1(x) and M2(x) be respectively two unknown regbression functions (response function with levelx). Three kinds of problems are considered.   (1) For given α, let θ1, and θ2 be respectively the unique root of M1 and M2, I.e. M1(θ1)=M2(θ2)=α. By taking samples, we want to decideθ1≧θ2 or θ1<θ2.   (2) For given α, we propose an algorithm to find a confidence interval for θ, with a prefixed confidence probability, where M1(θ)=α.   (3) Assume a unique θ such that M1(θ) attains its maximum, we propose an algorithm to find a confidence interval for θ with a prefixed confidence probability.   By using computer, some Monte Carlo Studies are made for the proposed algorithm.
author2 Huang, Wen-Tao
author_facet Huang, Wen-Tao
Su, Yu-Wen
蘇郁文
author Su, Yu-Wen
蘇郁文
spellingShingle Su, Yu-Wen
蘇郁文
On Finite Stages Stochastic Approximation with Some Empirical Studies
author_sort Su, Yu-Wen
title On Finite Stages Stochastic Approximation with Some Empirical Studies
title_short On Finite Stages Stochastic Approximation with Some Empirical Studies
title_full On Finite Stages Stochastic Approximation with Some Empirical Studies
title_fullStr On Finite Stages Stochastic Approximation with Some Empirical Studies
title_full_unstemmed On Finite Stages Stochastic Approximation with Some Empirical Studies
title_sort on finite stages stochastic approximation with some empirical studies
url http://ndltd.ncl.edu.tw/handle/97377841164907777274
work_keys_str_mv AT suyuwen onfinitestagesstochasticapproximationwithsomeempiricalstudies
AT sūyùwén onfinitestagesstochasticapproximationwithsomeempiricalstudies
AT suyuwen yǒuxiànbùzhòudesuíjībījìnjíqímùnǐyánjiū
AT sūyùwén yǒuxiànbùzhòudesuíjībījìnjíqímùnǐyánjiū
_version_ 1718188451445080064