A Multifaceted Quantification of Bias in Large Language Models

Simple item page
dc.contributor.advisorDaumé III, Halen_US
dc.contributor.authorSotnikova, Annaen_US
dc.contributor.departmentApplied Mathematics and Scientific Computationen_US
dc.contributor.publisherDigital Repository at the University of Marylanden_US
dc.contributor.publisherUniversity of Maryland (College Park, Md.)en_US
dc.date.accessioned2024-02-14T06:35:43Z
dc.date.available2024-02-14T06:35:43Z
dc.date.issued2023en_US
dc.description.abstractLanguage models are rapidly developing, demonstrating impressive capabilities in comprehending, generating, and manipulating text. As they advance, they unlock diverse applications across various domains and become increasingly integrated into our daily lives. Nevertheless, these models, trained on vast and unfiltered datasets, come with a range of potential drawbacks and ethical issues. One significant concern is the potential amplification of biases present in the training data, generating stereotypes and reinforcing societal injustices when language models are deployed. In this work, we propose methods to quantify biases in large language models. We examine stereotypical associations for a wide variety of social groups characterized by both single and intersectional identities. Additionally, we propose a framework for measuring stereotype leakage across different languages within multilingual large language models. Finally, we introduce an algorithm that allows us to optimize human data collection in conditions of high levels of human disagreement.en_US
dc.identifierhttps://doi.org/10.13016/0eex-wnsj
dc.identifier.urihttp://hdl.handle.net/1903/31722
dc.language.isoenen_US
dc.subject.pqcontrolledApplied mathematicsen_US
dc.subject.pqcontrolledComputer scienceen_US
dc.subject.pquncontrolledArtificial Intelligenceen_US
dc.subject.pquncontrolledEthicsen_US
dc.subject.pquncontrolledLarge Language Modelsen_US
dc.subject.pquncontrolledNatural Language Processingen_US
dc.subject.pquncontrolledStereotypesen_US
dc.titleA Multifaceted Quantification of Bias in Large Language Modelsen_US
dc.typeDissertationen_US

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Sotnikova_umd_0117E_23859.pdf
Size:
2.83 MB
Format:
Adobe Portable Document Format
Download

Collections

UMD Theses and Dissertations
Computer Science Theses and Dissertations
Mathematics Theses and Dissertations