Towards Few-shot Entity Recognition in Document Images: A Label-aware Sequence-to-Sequence Framework

Anonymous

Abstract: Entity recognition is a fundamental task in understanding document images. Traditional sequence labeling framework requires extensive datasets and high-quality annotations, which are typically expensive in practice. In this paper, we aim to build an entity recognition model based on only a few shots of annotated document images. To overcome the data limitation, we propose to leverage the label surface names to better inform the model of the target entity semantics. Specifically, we go beyond sequence labeling and develop a novel label-aware seq2seq framework, LASER. We design a new labeling scheme that generates the label surface names word-by-word explicitly after generating the entities. Moreover, we design special layout identifiers to capture the spatial correspondence between regions and labels. During training, LASER refines the label semantics by updating the label surface name representations and also strengthens the label-region correlation. In this way, LASER recognizes the entities from document images through both semantic and layout correspondence. Extensive experiments on two benchmark datasets demonstrate the superiority of LASER under the few-shot setting.