Seminars and Events

ISI Natural Language Seminar

Anti-Queer Bias in Large Language Models

Event Details

REMINDER:

Meeting hosts only admit guests that they know to the Zoom meeting. Hence, you’re highly encouraged to use your USC account to sign into Zoom.

If you’re an outside visitor, please inform us at (nlg-seminar-host(at)isi.edu) beforehand so we’ll be aware of your attendance and let you in.

In-person attendance will be permitted for USC/ISI faculty, staff, students only. Open to the public virtually via the zoom registration link and online.

For more information on the NL Seminar series and upcoming talks, please visit:

https://nlg.isi.edu/nl-seminar/ 

Happy Pride! To close out Pride Month at ISI, this talk will discuss fairness and bias in LLMs as it relates to the LGBTQ+ community. We will explore current methods for detecting and mitigating bias in LLMs, as well as the (lack of) current research focusing specifically on homophobic and transphobic biases. The talk will present recent exploratory work on whether and to what extent biases against queer and trans people are encoded in large language models (LLMs) such as BERT. It will discuss a new method for reducing these biases in downstream tasks: fine-tuning the models on data written by and/or about queer people. It will also discuss a new benchmark dataset, WinoQueer, modeled after other bias-detection benchmarks but addressing homophobic and transphobic biases. This work was accepted to the Queer in AI workshop at NAACL 2022.

Speaker Bio

Katy Felkner is a rising 3rd year PhD student at USC Information Sciences Institute. Her primary research focus is extremely low-resource machine translation. She is also interest in fairness and bias in large language models. Prior to USC, she received dual bachelor’s degrees in Computer Science and Letters (general humanities) from the University of Oklahoma. Her research is supported by an NSF Graduate Research Fellowship. Katy is passionate about making computer science more welcoming for women and queer students.

The recording for this AI Seminar talk will be posted on our USC/ISI YouTube page within 1-2 business days: https://www.youtube.com/user/USCISI.