The proposed project aims to develop new methodologies for developing NMT systems between extremely low-resource languages and English. Recent advances in neural machine translation (NMT) are a significant step forward in machine translation capabilities. However, "NMT systems have a steeper learning curve with respect to the amount of training data, resulting in worse quality in low-resource settings". A number of emerging approaches, such as zero resource and unsupervised NMT, have investigated alternative methods in developing NMT models where sufficient parallel corpora are not available (eg [1,2]). This project investigates methods to enable high performing NMT in low-resource scenarios.
 Neural Machine Translation for Bilingually Scarce Scenarios: A Deep Multi-task Learning Approach
Poorya Zaremoodi, Gholamreza Haffari
Proceedings of the 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT), 2018.
 Learning How to Actively Learn: A Deep Imitation Learning Approach
Ming Liu, Wray Buntine, Gholamreza Haffari
Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (ACL), 2018.