2 papers with code • 0 benchmarks • 0 datasets
The average of the normalized top-1 prediction scores of unseen classes in the generalized zero-shot learning setting, where the label of a test sample is predicted among all (seen + unseen) classes.
Many approaches in generalized zero-shot learning rely on cross-modal mapping between the image feature space and the class embedding space.
Ranked #1 on Generalized Few-Shot Learning on AWA2
In contrast, we propose a generative model that can naturally learn from unsupervised examples, and synthesize training examples for unseen classes purely based on their class embeddings, and therefore, reduce the zero-shot learning problem into a supervised classification task.