"Wikily" Supervised Neural Translation Tailored to Cross-Lingual Tasks

We present a simple but effective approach for leveraging Wikipedia for neural machine translation as well as cross-lingual tasks of image captioning and dependency parsing without using any direct supervision from external parallel data or supervised models in the target language. We show that first sentences and titles of linked Wikipedia pages, as well as crosslingual image captions, are strong signals for a seed parallel data to extract bilingual dictionaries and cross-lingual word embeddings for mining parallel text from Wikipedia. Our final model achieves high BLEU scores that are close to or sometimes higher than strong supervised baselines in low-resource languages; e.g. supervised BLEU of 4.0 versus 12.1 from our model in English-to-Kazakh. Moreover, we tailor our "wikily" supervised translation models to unsupervised image captioning, and cross-lingual dependency parser transfer. In image captioning, we train a multitasking machine translation and image captioning pipeline for Arabic and English from which the Arabic training data is a translated version of the English captioning data, using our wikily-supervised translation models. Our captioning results on Arabic are slightly better than that of its supervised model. In dependency parsing, we translate a large amount of monolingual text, and use it as artificial training data in an annotation projection framework. We show that our model outperforms recent work on cross-lingual transfer of dependency parsers. * Research was conducted at The University of Pennsylvania. 117 lingual data for machine translation without 118 any explicit supervision. Our mining algo-119 rithm easily scales on large comparable data 120 using limited computational resources. We 121 achieve very high BLEU scores for distant 122 languages, especially those in which current 123 unsupervised methods perform very poorly. 124 We propose novel methods for leveraging 125 our current translation models in image cap-126 tioning. We show that how a combina-127 tion of translating caption training data, and 128 multi-task learning with English captioning as 129 well as translation improves the performance. 130 Our results on Arabic captaining shows re-131 sults slightly superior to that of a supervised 132 captioning model trained on gold-standard 133 datasets. 134 We propose a novel modification to the anno-135 tation projection method in order to be able 136 to leverage our translation models. Our re-137 sults on dependency parsing performs better 138 than previous work in most cases, and per-139 forms similarly to using gold-standard parallel 140 datasets.