Zero-shot learning is an advanced machine - learning technique that empowers a model to identify and categorize novel objects without having been trained on specific examples of those objects. This remarkable feat is accomplished through a knowledge transfer method. In this process, the model is provided with information regarding the relationships among different classes. Leveraging this knowledge, the model can then make predictions about objects it has never encountered before.
