Zero-Shot Learning

Prototype networks

Neural network architecture that learns a metric space where classification is performed by computing distances to class prototypes, which are mean representatives of each category.

← Tillbaka