Model-Agnostic Meta Learning for Class Imbalance Adaptation
arXiv:2604.18759v1 Announce Type: new
Abstract: Class imbalance is a widespread challenge in NLP tasks, significantly hindering robust performance across diverse domains and applications. We introduce Hardness-Aware Meta-Resample (HAMR), a unified fra…