Unified Pretraining Framework for Document Understanding
This work addresses the need for reduced annotation efforts in document intelligence for business applications, though it is incremental as it builds on existing self-supervised learning methods.
The paper tackles the problem of language-dominated pretraining in document understanding by introducing UDoc, a unified multimodal pretraining framework that uses three self-supervised losses, resulting in improved joint representations and downstream task performance.
Document intelligence automates the extraction of information from documents and supports many business applications. Recent self-supervised learning methods on large-scale unlabeled document datasets have opened up promising directions towards reducing annotation efforts by training models with self-supervised objectives. However, most of the existing document pretraining methods are still language-dominated. We present UDoc, a new unified pretraining framework for document understanding. UDoc is designed to support most document understanding tasks, extending the Transformer to take multimodal embeddings as input. Each input element is composed of words and visual features from a semantic region of the input document image. An important feature of UDoc is that it learns a generic representation by making use of three self-supervised losses, encouraging the representation to model sentences, learn similarities, and align modalities. Extensive empirical analysis demonstrates that the pretraining procedure learns better joint representations and leads to improvements in downstream tasks.