Show simple item record

dc.contributor.authorBlum, Sophie Martina
dc.date.accessioned2023-08-29T23:39:53Z
dc.date.available2023-08-29T23:39:53Z
dc.date.issued2023-08-01
dc.date.submitted2023-08-29T22:00:23Z
dc.identifier.urihttps://hdl.handle.net/11250/3086287
dc.description.abstractWe investigate an approach for extracting occupational gender bias in the form of logical rules from Large Language Models (LLM)s based on Angluin's exact learning model with membership and equivalence queries to an oracle. In our approach, the oracle is a LLM and we show the changes that are necessary to use Angluin's algorithm with such an oracle. In our experiments, we extract occupational gender bias with the adapted algorithm from BERT and roBERTa models and compare our results to an established bias extraction method, which is template-based probing. Our goal is to use a new method to combine multiple attributes in a template sentence and to study their relationship to the gender in a sentence. We achieve this by using our rule extraction approach with a variable template containing multiple attributes. The extracted rules show a similar bias as previous bias extraction methods but also give insight into more complex relationships between attributes.
dc.language.isoeng
dc.publisherThe University of Bergen
dc.rightsCopyright the Author. All rights reserved
dc.subjectexact learning
dc.subjectnatural language processing
dc.subjectmachine learning
dc.titleInvestigating Biases in Rules Extracted from Language Models
dc.typeMaster thesis
dc.date.updated2023-08-29T22:00:23Z
dc.rights.holderCopyright the Author. All rights reserved
dc.description.degreeMaster's Thesis in Informatics
dc.description.localcodeINF399
dc.description.localcodeMAMN-PROG
dc.description.localcodeMAMN-INF
dc.subject.nus754199
fs.subjectcodeINF399
fs.unitcode12-12-0


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record