Improving Few-Shot Named Entity Recognition with Causal Interventions
Few-shot Named Entity Recognition (NER) systems are designed to identify new categories of entities with a limited number of labeled examples. A major challenge encountered by these systems is overfitting, particularly pronounced in comparison to tasks with ample samples. This overfitting predominan...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Tsinghua University Press
2024-12-01
|
| Series: | Big Data Mining and Analytics |
| Subjects: | |
| Online Access: | https://www.sciopen.com/article/10.26599/BDMA.2024.9020052 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|