"This talk will cover our recent work on developing deep learning algorithms with applications to biomedical narrative text. The common theme of these studies aims at building models that improve prediction accuracy by exploring and combining relational information in text.
The talk will focus on concrete examples including biomedical relation extraction (short text understanding) from clinical notes and document classification (long text understanding). In each example, Dr. Luo will show how to automatically build relational information into a graph representation and how to learn features from graphs. Depending on the nature of the task, heavier machinery of convolutional or recurrent neural networks become necessary to reliably capture important features. Dr. Luo will demonstrate that these methods lead to progressively improved performance by integrating lexical, syntactic and semantic information."
NOTE: Free to AMIA members, $50 for non-members