
Flattening representation-space loss landscapes improves cross-domain few-shot learning by enhancing transferability and fine-tuning.


coremsg

cross-domain-few-shot-learning-with-flattened-loss-landscapes


Cross-Domain Few-Shot Learning with Flattened Loss Landscapes



Flattening long-range loss landscapes in the representation space enhances transferability and fine-tuning in cross-domain few-shot learning.



Cross-Domain Few-Shot Learning: Flattening Long-Range Loss Landscapes
