Friday, April 9 | 10 to 11 a.m. | Zoom Meeting

Please join us for the Informatics Institute PowerTalk Seminar Series-Bioinformatics!

“Beyond BERT: Learning Representations for Clinical NLP”

Timothy Miller, Ph.D.
Assistant Professor of Pediatrics and Computational Health Informatics Program (CHIP)
Harvard Medical School

Abstract: Pre-trained language models like Bidirectional Encoder Representations from Transformers (BERT) have revolutionized naatural language processing (NLP), including NLP on biomedical texts like electronic health record data. While offering improved accuracy, these models have weaknesses and quirks that offer unique opportunities for novel research directions. In this talk, I describe three recent works from my lab where we have taken advantage of the power of BERT-style models and extended their possibilities in creative new directions to make progress on important research questions.

Register for Zoom Meeting

Click here to register for the seminar. Once you have registered you will receive a confirmation email containing information on how to access the webinar.