Unpaired Cross-lingual Image Caption Generation With Self-supervised Rewards

Song Yuqing, Chen Shizhe, Zhao Yida, Jin Qin. Arxiv 2019

[Paper]    
Applications Reinforcement Learning Tools Training Techniques

Generating image descriptions in different languages is essential to satisfy users worldwide. However, it is prohibitively expensive to collect large-scale paired image-caption dataset for every target language which is critical for training descent image captioning models. Previous works tackle the unpaired cross-lingual image captioning problem through a pivot language, which is with the help of paired image-caption data in the pivot language and pivot-to-target machine translation models. However, such language-pivoted approach suffers from inaccuracy brought by the pivot-to-target translation, including disfluency and visual irrelevancy errors. In this paper, we propose to generate cross-lingual image captions with self-supervised rewards in the reinforcement learning framework to alleviate these two types of errors. We employ self-supervision from mono-lingual corpus in the target language to provide fluency reward, and propose a multi-level visual semantic matching model to provide both sentence-level and concept-level visual relevancy rewards. We conduct extensive experiments for unpaired cross-lingual image captioning in both English and Chinese respectively on two widely used image caption corpora. The proposed approach achieves significant performance improvement over state-of-the-art methods.

Similar Work