Target-Bidirectional Neural Models for Machine Transliteration

作者：Andrew Finch, Lemao Liu, Xiaolin Wang, Eiichiro Sumita · 年份：2016 · DOI：10.18653/v1/w16-2711 · 被引用次数：32 · 研究领域：Handwritten Text Recognition Techniques、Natural Language Processing Techniques、Image and Object Detection Techniques

Our purely neural network-based system represents a paradigm shift away from the techniques based on phrase-based statistical machine translation we have used in the past. The approach exploits the agreement between a pair of target-bidirectional LSTMs, in order to generate balanced targets with both good suffixes and good prefixes. The evaluation results show that the method is able to match and even surpass the current state-of-the-art on most language pairs, but also exposes weaknesses on some tasks motivating further study. The Janus toolkit that was used to build the systems used in the evaluation is publicly available at https://github.com/lemaoliu/Agtarbidir.