What is Zero-shot Learning?
It is the ability of artificial intelligence to predict something for which it has no training.
Overview
Zero-sample learning is the ability of artificial intelligence to make a prediction on a subject for which it has never been trained or seen an example before. Thanks to its general knowledge of the world, the model solves a task it has never seen by reasoning.
How it works
The model uses conceptual relationships learned during its training. For example, if he knows the concepts of 'cat' and 'dog', he can identify a type of animal he has never seen when it is described.
Where it is used
It is frequently used in casual chatbots, where users ask questions directly to the model without giving any examples.
Commonly confused with
It is confused with few-shot learning; Here the model has no clues or examples.
Frequently asked questions
How can the model make predictions?
He reasons by using the similarities between concepts in the huge data pool he learned during his education.
Is zero-shot always accurate?
No, there is a higher risk of hallucinating things that are beyond the model's knowledge.
Related terms
This explanation was written in plain language for TreScout and machine-translated from the Turkish original · the Turkish version prevails. If something looks wrong or missing, write to hello@trescout.com. Read in Turkish →