Submitted by Anonymous (not verified) on Wed, 11/16/2022 - 11:19
Abstract
The availability of large multilingual pre-trained language models has opened up exciting pathways for developing NLP technologies for languages with scarce resources. In this talk I will advocate for the need to go beyond the most common languages in multilingual evaluation, and on the challenges of handling new, unseen-during-training languages and varieties. I will also share some of my experiences with working with indigenous and other endangered language communities and activists.
Biography
Submitted by Anonymous (not verified) on Wed, 11/16/2022 - 11:19
Abstract
I will present our work on data augmentation using style transfer as a way to improve domain adaptation in sequence labeling tasks. The target domain is social media data, and the task is named entity recognition (NER). The premise is that we can transform the labelled out of domain data into something that stylistically is more closely related to the target data. Then we can train a model on a combination of the generated data and the smaller amount of in domain data to improve NER prediction performance. I will show recent empirical results on these efforts.
You can help by contributing your favorite holiday dish (regional specialties strongly encouraged!) to this pot-luck get together (you don’t have to bring anything to participate). Main dishes will be provided, as will plates, napkins, utensils, etc. Click here to sign up
There will a gingerbread decorating contest and prizes for best/ugliest sweater!
Submitted by Anonymous (not verified) on Wed, 11/16/2022 - 11:19
Link for Live Seminar
Link for Recorded seminars – 2022/2023 school year
Panel Speaker 1: Erin Sutton, PhD
Guidance and Control Engineer at the JHU Applied Physics Laboratory
Ph.D. Mechanical Engineering 2017, M.S. Mechanical Engineering 2016