Zero-shot Classification

Zero-shot classification is a type of machine learning task where a model learns to recognize classes it hasn't seen before during training. Unlike typical classification tasks where all classes are known upfront, zero-shot classification allows a model to generalize to new classes using semantic descriptions or attributes associated with each class. This method is valuable when dealing with a large or changing number of possible classes, making it impractical to have labeled training data for every class.

‍

‍