EVENT DETAILS
In this talk, I will cover two of the recent papers in our group on applying recurrent and convolutional neural networks to Natural Language Processing. Automated extraction of knowledge from biomedical literature or clinical notes involves accurately identifying, not only the conceptual entities, but also the varied relationships among those concepts. Relation identification has received increasing attention over the past decade, and is critical in applications including clinical decision making, clinical trial screening, and pharmacovigilance. We propose Segment Long Short-Term Memory networks (Seg-LSTM) and Segment Convolutional Neural Networks (Seg-CNN) for classifying relations from clinical notes. Both Seg-LSTM and Seg-CNN achieved state-of-the-art performance using only word embedding features without manual feature engineering. We expect much can be done in this longstanding difficult NLP field and beyond with the new technique.
TIME Tuesday October 17, 2017 at 12:00 PM - 1:00 PM
LOCATION Lakeview Conference Room (11th Floor), Arthur Rubloff Building map it
ADD TO CALENDAR&group= echo $value['group_name']; ?>&location= echo htmlentities($value['location']); ?>&pipurl= echo $value['ppurl']; ?>" class="button_outlook_export">
CONTACT Lindsay Varasteh lindsay.varasteh1@northwestern.edu
CALENDAR Center for Data Science and Informatics (CDSI)