{"379491":{"#nid":"379491","#data":{"type":"event","title":"Ph.D Defense by Huizhu (Crystal) Wang","body":[{"value":"\u003Cp\u003E*Title:*\u003Cbr \/\u003E\u003Cstrong\u003EStatistical Selection and Wavelet-Based Profile Monitoring\u003C\/strong\u003E\u003Cbr \/\u003E\u003Cbr \/\u003E\u003Cstrong\u003E*Advisors: *\u003C\/strong\u003E\u003Cbr \/\u003EDr. Seong-Hee Kim\u003Cbr \/\u003E\u003Cbr \/\u003E\u003Cstrong\u003E*Committee:*\u003C\/strong\u003E\u003Cbr \/\u003EDr. Xiaoming Huo, Dr. Jianjun Shi, Dr. James R. Wilson (North Carolina\u003Cbr \/\u003EState University), and Dr. Youngmi Hur (Yonsei University)\u003Cbr \/\u003E\u003Cbr \/\u003E\u003Cstrong\u003E*Date and time:*\u003C\/strong\u003E\u003Cbr \/\u003EThursday, February 26 2015, 9:30AM\u003Cbr \/\u003E\u003Cbr \/\u003E\u003Cstrong\u003E*Location:\u0026nbsp; *\u003C\/strong\u003E\u003Cbr \/\u003EAcademic Office - Groseclose 204\u003Cbr \/\u003E\u003Cbr \/\u003E\u003Cstrong\u003E*Abstract:*\u003C\/strong\u003E\u003Cbr \/\u003E\u003Cbr \/\u003EThis thesis consists of two topics: statistical selection and profile\u003Cbr \/\u003Emonitoring. Statistical selection is related to ranking and selection in\u003Cbr \/\u003Esimulation and profile monitoring is related to statistical process control.\u003Cbr \/\u003E\u003Cbr \/\u003ERanking and selection (R\u0026amp;S) is to select a system with the largest or\u003Cbr \/\u003Esmallest performance measure among a finite number of simulated\u003Cbr \/\u003Ealternatives with some guarantee about correctness. Fully sequential\u003Cbr \/\u003Eprocedures have been shown to be efficient, but their actual probabilities\u003Cbr \/\u003Eof correct selection tend to be higher than the nominal level, implying\u003Cbr \/\u003Ethat they consume unnecessary observations. In the first part, we study\u003Cbr \/\u003Ethree conservativeness sources in fully sequential indifference-zone (IZ)\u003Cbr \/\u003Eprocedures and use experiments to quantify the impact of each source in\u003Cbr \/\u003Eterms of the number of observations, followed by an asymptotic analysis on\u003Cbr \/\u003Ethe impact of the critical one. Then we propose new asymptotically valid\u003Cbr \/\u003Eprocedures that lessen the critical conservativeness source, by mean update\u003Cbr \/\u003Ewith or without variance update. Experimental results showed that new\u003Cbr \/\u003Eprocedures achieved meaningful improvement on the efficiency.\u003Cbr \/\u003E\u003Cbr \/\u003EThe second part is developing a wavelet-based distribution-free tabular\u003Cbr \/\u003ECUSUM chart based on adaptive thresholding. WDFTCa is designed for rapidly\u003Cbr \/\u003Edetecting shifts in the mean of a high-dimensional profile whose noise\u003Cbr \/\u003Ecomponents have a continuous nonsingular multivariate distribution. First\u003Cbr \/\u003Ecomputing a discrete wavelet transform of the noise vectors for randomly\u003Cbr \/\u003Esampled Phase I (in-control) profiles, WDFTCa uses a matrix-regularization\u003Cbr \/\u003Emethod to estimate the covariance matrix of the wavelet-transformed noise\u003Cbr \/\u003Evectors; then those vectors are aggregated (batched) so that the\u003Cbr \/\u003Enonoverlapping batch means of the wavelet-transformed noise vectors have\u003Cbr \/\u003Emanageable covariances. Lower and upper in-control thresholds are computed\u003Cbr \/\u003Efor the resulting batch means of the wavelet-transformed noise vectors\u003Cbr \/\u003Eusing the associated marginal Cornish-Fisher expansions that have been\u003Cbr \/\u003Esuitably adjusted for between-component correlations. From the thresholded\u003Cbr \/\u003Ebatch means of the wavelet-transformed noise vectors, Hotelling\u2019s T^2-type\u003Cbr \/\u003Estatistics are computed to set the parameters of a CUSUM procedure. To\u003Cbr \/\u003Emonitor shifts in the mean profile during Phase II (regular) operation,\u003Cbr \/\u003EWDFTCa computes a similar Hotelling\u2019s T^2-type statistic from successive\u003Cbr \/\u003Ethresholded batch means of the wavelet-transformed noise vectors using the\u003Cbr \/\u003Ein-control thresholds; then WDFTCa applies the CUSUM procedure to the\u003Cbr \/\u003Eresulting T^2-type statistics. Experimentation with several normal and\u003Cbr \/\u003Enonnormal test processes revealed that WDFTCa outperformed existing\u003Cbr \/\u003Enonadaptive profile-monitoring schemes.\u003Cbr \/\u003E\u003Cbr \/\u003E\u003C\/p\u003E","summary":null,"format":"limited_html"}],"field_subtitle":"","field_summary":"","field_summary_sentence":[{"value":"Statistical Selection and Wavelet-Based Profile Monitoring"}],"uid":"27707","created_gmt":"2015-02-18 13:28:19","changed_gmt":"2016-10-08 01:47:42","author":"Tatianna Richardson","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2015-02-26T08:30:00-05:00","event_time_end":"2015-02-26T10:30:00-05:00","event_time_end_last":"2015-02-26T10:30:00-05:00","gmt_time_start":"2015-02-26 13:30:00","gmt_time_end":"2015-02-26 15:30:00","gmt_time_end_last":"2015-02-26 15:30:00","rrule":null,"timezone":"America\/New_York"},"extras":[],"groups":[{"id":"221981","name":"Graduate Studies"}],"categories":[],"keywords":[{"id":"1366","name":"defense"},{"id":"1808","name":"graduate students"},{"id":"105701","name":"Ph.d. Defense"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[{"id":"1788","name":"Other\/Miscellaneous"}],"invited_audience":[{"id":"78771","name":"Public"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}