2026-09-18T22:35:34Zhttps://dspace.mit.edu/server/oai/request

oai:dspace.mit.edu:1721.1/1203732026-06-16T18:54:28Zcom_1721.1_7582com_1721.1_7581col_1721.1_131022

Constantinos Daskalakis. Kamath, Gautam (Gautam Chetan) Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science. Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science 2019-02-14T15:22:13Z 2019-02-14T15:22:13Z 2018 2018 http://hdl.handle.net/1721.1/120373 1084273332 Thesis: Ph. D., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2018. This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. Cataloged from student-submitted PDF version of thesis. Includes bibliographical references (pages 321-356). Hypothesis testing is one of the most classical problems in statistics. While it has enjoyed over a century of intense study, only recent focus has been on the small-sample regime, with interest in sample complexities and minimax rates. Our understanding of many fundamental problems is now quite mature, but there are several questions which have arisen over the last decade, which have not yet received adequate attention. The goal of this dissertation is to identify and address several contemporary challenges in distribution testing. In particular, we make progress in answering the following questions: ** Can we test distributions with tolerance to model misspecification? ** How does the complexity of distribution testing change as we consider different measures of distance? ** Can we efficiently test for membership in (potentially infinite) classes of distributions? ** How can we avoid the curse of dimensionality when testing multivariate distributions? ** Is it possible to perform hypothesis testing on sensitive data, while respecting privacy of the dataset? ** Can we design more efficient algorithms if the dataset is sampled actively? Directions for further investigation are also discussed. by Gautam Kamath. Ph.D. 356 pages eng Massachusetts Institute of Technology MIT theses are protected by copyright. They may be viewed, downloaded, or printed from this source but further reproduction or distribution in any format is prohibited without written permission. http://dspace.mit.edu/handle/1721.1/7582 Electrical Engineering and Computer Science. Modern challenges in distribution testing Thesis application/pdf false Publication unknown unknown <Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="3ce93ba3-5dc9-4485-9c5f-952e4d08fbc4"> <Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type> <Language>eng</Language> <Title>Modern challenges in distribution testing</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2018</PublicationDate> <Authors> <Author> <DisplayName>Kamath, Gautam (Gautam Chetan)</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <DisplayName>Massachusetts Institute of Technology</DisplayName> <OrgUnit /> </Publisher> </Publishers> <License>http://dspace.mit.edu/handle/1721.1/7582</License> <Keyword>Electrical Engineering and Computer Science.</Keyword> <Abstract>Hypothesis testing is one of the most classical problems in statistics. While it has enjoyed over a century of intense study, only recent focus has been on the small-sample regime, with interest in sample complexities and minimax rates. Our understanding of many fundamental problems is now quite mature, but there are several questions which have arisen over the last decade, which have not yet received adequate attention. The goal of this dissertation is to identify and address several contemporary challenges in distribution testing. In particular, we make progress in answering the following questions: ** Can we test distributions with tolerance to model misspecification? ** How does the complexity of distribution testing change as we consider different measures of distance? ** Can we efficiently test for membership in (potentially infinite) classes of distributions? ** How can we avoid the curse of dimensionality when testing multivariate distributions? ** Is it possible to perform hypothesis testing on sensitive data, while respecting privacy of the dataset? ** Can we design more efficient algorithms if the dataset is sampled actively? Directions for further investigation are also discussed.</Abstract> <Access xmlns="http://purl.org/coar/access_right" > </Access> </Publication>