Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models
Keywords:Speech & Natural Language Processing (SNLP), Machine Learning (ML)
AbstractWe investigate the use of multimodal information contained in images as an effective method for enhancing the commonsense of Transformer models for text generation. We perform experiments using BART and T5 on concept-to-text generation, specifically the task of generative commonsense reasoning, or CommonGen. We call our approach VisCTG: Visually Grounded Concept-to-Text Generation. VisCTG involves captioning images representing appropriate everyday scenarios, and using these captions to enrich and steer the generation process. Comprehensive evaluation and analysis demonstrate that VisCTG noticeably improves model performance while successfully addressing several issues of the baseline generations, including poor commonsense, fluency, and specificity.
How to Cite
Feng, S. Y., Lu, K., Tao, Z., Alikhani, M., Mitamura, T., Hovy, E., & Gangal, V. (2022). Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models. Proceedings of the AAAI Conference on Artificial Intelligence, 36(10), 10618-10626. https://doi.org/10.1609/aaai.v36i10.21306
AAAI Technical Track on Speech and Natural Language Processing