Teaching machines to understand human language is one of the most elusive and long-standing challenges in Natural Language Processing (NLP). Driven by the fast development of deep learning, state-of-the-art NLP models have already achieved human-level performance in various large benchmark datasets, such as SQuAD, SNLI, and RACE. However, when we try to apply these strong models to real-world applications, we are still facing two challenges: 1. We do not have enough data samples for model training for a specific problem; 2. Deployed models may crash on noisy test data or natural/artificial adversaries. In short, performance degradation on low-resource tasks/datasets and low robustness are the two obstacles to prevent NLP models from being massively applied into the wild. This dissertation aims to address these two issues. Towards the first one, we resort to transfer learning to leverage knowledge acquired from related data in order to improve performance on a target low-resource task/dataset. For the second issue, we propose methods to evaluate the robustness of NLP models on text classification and entailment tasks.
Committee: Peter Szolovits (Advisor), John J. Leonard (Chair), Jim Glass, Sanjay Sarma, Brian Subirana
To attend this defense, please contact this MechE doctoral candidate at jindi15 at mit dot edu