Look, Read and Enrich. Learning from Scientific Figures and their Captions
This work addresses the problem of machine comprehension of scientific figures for researchers and AI systems, representing an incremental advance by applying unsupervised learning to a previously untapped data source.
The paper tackles the challenge of understanding scientific figures by leveraging the correspondence between figures and their captions, introducing an unsupervised learning task that enriches features with knowledge graphs and improves performance on multi-modal classification and question answering tasks.
Compared to natural images, understanding scientific figures is particularly hard for machines. However, there is a valuable source of information in scientific literature that until now has remained untapped: the correspondence between a figure and its caption. In this paper we investigate what can be learnt by looking at a large number of figures and reading their captions, and introduce a figure-caption correspondence learning task that makes use of our observations. Training visual and language networks without supervision other than pairs of unconstrained figures and captions is shown to successfully solve this task. We also show that transferring lexical and semantic knowledge from a knowledge graph significantly enriches the resulting features. Finally, we demonstrate the positive impact of such features in other tasks involving scientific text and figures, like multi-modal classification and machine comprehension for question answering, outperforming supervised baselines and ad-hoc approaches.