zero-shot learning en · NOUN
Meanings
- (qualifier:machine learning, uncountable) A problem setup in machine learning, where, at test time, a learner observes samples from classes that were not observed during training and attempts to predict the class that they belong to, typically based on auxiliary information.