Attribute-Based Learning

Semantic Projection Bias

Phenomenon where ZSL models tend to preferentially predict classes seen during training, even in generalized zero-shot scenarios.

← Kembali