Font Size: a A A

Parameter Estimation Of Two-stage Adaptive Group Testing In Gaussian Mixture Model

Posted on:2020-09-28Degree:MasterType:Thesis
Country:ChinaCandidate:L JiangFull Text:PDF
GTID:2417330596974388Subject:Applied statistics
Abstract/Summary:PDF Full Text Request
Grouping detection is widely used in various industries,which has the advantages of reducing cost,improving efficiency and accuracy.At the beginning,grouping detection took samples subject to the 0-1 binomial distribution as the research object,and the group size was usually a fixed constant value in the grouping process.Until later,some scholars proposed that the selection of group size should be optimized to improve the accuracy of the grouping detection results,adaptive grouping detection appeared.Parameter estimation of adaptive grouping detection is based on grouping detection in stages,and group size is constantly updated from one stage to the next.Two points are based on the size of the alternation group.The first is to use the data obtained in the previous stage to carry out maximum likelihood estimation.The second is that the number of groups detected will change in the next stage.Therefore,the1N number of groups in the first stage is group,and there are K2 individuals in each group;theN2 number of groups in the second stage is group,and there are1K individuals in each group,and theN3 number of groups in the next stage is group,and there areK3 individuals in each group from this.Under the binomial distribution,the size of the adaptive optimal group is determined by the number of groups and the number of groups.The main content of this paper is to expand the sample from the discrete binomial distribution to the continuous gaussian distribution for two-stage adaptive grouping detection.Since the traditional EM algorithm cannot solve the analytical solution,this paper first introduces the gaussian mixture model under limited conditions and the basic definition of grouping detection in disease detection,and then calculates the size of the optimal group K,where the size of the optimal group K is related to the five restricted parameters to be estimated.At the same time,the two-stage adaptive calculation process is described,and the two-stage parameter estimation is carried out by using EM algorithm in the constrained gaussian mixture model.The calculation steps are divided into three steps.Finally,numerical simulation is carried out with the algorithm,and the results show that compared with non-adaptive detection and monomer detection,the two-stage adaptive grouping detection statistics of gaussian mixture model have good statistical properties.
Keywords/Search Tags:Adaptive, Grouping detection, Two stage, Optimal group size, Gaussian mixture model
PDF Full Text Request
Related items