Natural language understanding with imitation learning

17^th May 2016, 13:00 Ashton Lecture Theater
Dr. Andreas Vlachos
University of Sheffield

Abstract

In this talk I will focus on the machine learning approach used in most of my work in natural language understanding, namely imitation learning. In this paradigm, structured prediction is converted into a sequence of classification actions which are learnt by appropriate generation of cost-sensitive training examples, so that the impact of the actions is assessed globally while maintaining flexibility in feature extraction. Experiments in a variety of tasks in biomedical text mining, information extraction and semantic parsing demonstrate the broad applicability and the competitive performance of imitation learning. Finally, time permitting, I will give an overview of my recent work in fact-checking and language modelling.