DataFunSummit

Jul 14, 2026 · Artificial Intelligence

Memory‑Guided Hard Data Augmentation: Turning Model Errors into Targeted Multimodal NER Improvements

The paper proposes Memory‑Guided Hard Data Augmentation (MGHDA), a closed‑loop pipeline that diagnoses model‑specific hard instances in multimodal named entity recognition, abstracts their error patterns into a Memory Tree, and generates targeted augmentation samples, achieving consistent F1 gains across several backbones while highlighting cost and scalability trade‑offs.