{"id":{"repo_id":"buffalo","oai_identifier":"oai:ubir.buffalo.edu:10477/84065"},"canonical_url":"https://search.dev.ndltd.org/etd/buffalo/oai:ubir.buffalo.edu:10477/84065","repository":{"repo_id":"buffalo","name":"Buffalo","base_url":"https://ubir.buffalo.edu/oai/request"},"display":{"title":"Nonparametric Quickest Change Detection for High Dimensional Sequential Data","abstract":"M.S.","abstract_html":"M.S.","abstract_has_math":false,"creators":["Wang, Jingtao; 0000-0002-2795-0546"],"institution":"State University of New York at Buffalo","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Zou, Shaofeng","Electrical Engineering"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2022,"date_issued":"2022-06-21T15:47:33Z","date_published":"2022-06-21T15:47:33Z","updated_at":"2026-07-27T19:05:30Z","subjects":["electrical engineering"],"languages":["eng"],"rights":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/10477/84065","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Zou, Shaofeng","Electrical Engineering"]},{"key":"dc:creator","label":"Author","values":["Wang, Jingtao; 0000-0002-2795-0546"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2022-06-21T15:47:33Z","2020"]},{"key":"dc:publisher","label":"Institution","values":["State University of New York at Buffalo"]},{"key":"dc:type","label":"Dc Type","values":["Text","Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["electrical engineering"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/10477/84065"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["M.S.","The problem of accurately and timely detecting abrupt changes in a stochastic system is of crucial importance in plenty of areas. It has lots of important applications, such as detecting the failures and intrusions in the cyber-physical system (CPS), e.g., power systems and Internet of Things networks. Among all of these applications, the distribution of the observations undergoes a change in response to an anomaly that occurs in the system, and the goal is to detect the change as quickly as possible subject to a false alarm constraint. Previous methods on anomaly detection use only a single observation to detect the anomaly, which may incur a high level of false alarms. Moreover, many approaches assume that we have some pre-knowledge of the pre-change or post-change distributions. However, in modern applications, the samples observed usually have a very high dimension, and thus estimating their distributions may be intractable in practice due to curse of dimensionality. In this thesis, the focus is to design new algorithms to address the above challenges. We focus on the problem of nonparametric quickest change detection for high dimensional sequential data. In our problem, the observations follow different distributions before and after the change point, and we assume that we don’t have any prior knowledge of the pre-change or post-change distributions. The goal is to detect the change as quickly as possible subject to a false alarm constraint. To solve our problem, we propose two generative model based quickest change detection algorithms. The first algorithm is a generative adversarial network (GAN) based CuSum-type algorithm, which uses the output of a pre-trained discriminator in GAN together with a CuSum-type detection statistic to perform quickest change detection. The second proposed algorithm is a variational autoencoder generative adversarial network (VAEGAN) based quickest change detection algorithm, which combines two generative models, VAE and GAN, the sliding window method and uses a linear score function to perform detection. We evaluate our proposed algorithms on a synthetic dataset and two real datasets. Compared with two previous algorithms, our proposed algorithms have the best performance.","**To request an accessible version of the file(s) associated with this item, contact library@buffalo.edu. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.**"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Nonparametric Quickest Change Detection for High Dimensional Sequential Data"]}]}],"canonical_facts":{"dc:contributor":["Zou, Shaofeng","Electrical Engineering"],"dc:creator":["Wang, Jingtao; 0000-0002-2795-0546"],"dc:date":["2022-06-21T15:47:33Z","2020"],"dc:description":["M.S.","The problem of accurately and timely detecting abrupt changes in a stochastic system is of crucial importance in plenty of areas. It has lots of important applications, such as detecting the failures and intrusions in the cyber-physical system (CPS), e.g., power systems and Internet of Things networks. Among all of these applications, the distribution of the observations undergoes a change in response to an anomaly that occurs in the system, and the goal is to detect the change as quickly as possible subject to a false alarm constraint. Previous methods on anomaly detection use only a single observation to detect the anomaly, which may incur a high level of false alarms. Moreover, many approaches assume that we have some pre-knowledge of the pre-change or post-change distributions. However, in modern applications, the samples observed usually have a very high dimension, and thus estimating their distributions may be intractable in practice due to curse of dimensionality. In this thesis, the focus is to design new algorithms to address the above challenges. We focus on the problem of nonparametric quickest change detection for high dimensional sequential data. In our problem, the observations follow different distributions before and after the change point, and we assume that we don’t have any prior knowledge of the pre-change or post-change distributions. The goal is to detect the change as quickly as possible subject to a false alarm constraint. To solve our problem, we propose two generative model based quickest change detection algorithms. The first algorithm is a generative adversarial network (GAN) based CuSum-type algorithm, which uses the output of a pre-trained discriminator in GAN together with a CuSum-type detection statistic to perform quickest change detection. The second proposed algorithm is a variational autoencoder generative adversarial network (VAEGAN) based quickest change detection algorithm, which combines two generative models, VAE and GAN, the sliding window method and uses a linear score function to perform detection. We evaluate our proposed algorithms on a synthetic dataset and two real datasets. Compared with two previous algorithms, our proposed algorithms have the best performance.","**To request an accessible version of the file(s) associated with this item, contact library@buffalo.edu. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.**"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/10477/84065"],"dc:language":["eng"],"dc:publisher":["State University of New York at Buffalo"],"dc:rights":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."],"dc:subject":["electrical engineering"],"dc:title":["Nonparametric Quickest Change Detection for High Dimensional Sequential Data"],"dc:type":["Text","Thesis"]},"updated_at":"2026-07-27T19:05:30Z"}