Papers by Nobuyuki Iokawa
Visual-Textual Entailment with Quantities Using Model Checking and Knowledge Injection (2024.lrec-main)
Copied to clipboard
| Challenge: | Visual-textual entailment (VTE) is a critical task in multimodal inference. |
| Approach: | They propose a visual-textual entailment system that solves VTE tasks with quantities and negation. |
| Outcome: | The proposed system solves visual-textual entailment tasks with quantities and negation more robustly than previous approaches. |