Papers with order-agnostic data augmentation
Order-Agnostic Data Augmentation for Few-Shot Named Entity Recognition (2024.acl-long)
Copied to clipboard
| Challenge: | Existing DA methods for named entity recognition (NER) are costly and labor-intensive to acquire, necessitating innovative approaches to data scarcity. |
| Approach: | They propose an order-agnostic data augmentation solution that exploits the order-based property in the training phase of sequence-to-sequence NER methods for data augmented. |
| Outcome: | The proposed method significantly enhances the few-shot capabilities of pre-trained language models in low-resource settings. |