Zero-shot Learning
An AI model performs a task it has never seen examples of, without needing any training samples for that specific job.
In Plain English
Zero-shot learning is when a trained model applies what it learned to completely new situations without any examples to guide it. Think of it like knowing the rules of chess so well that you could instantly understand how to play a variant you've never encountered. This is powerful because you don't have to collect and label new training data every time you want the model to do something slightly different. The model relies on general knowledge it picked up during training to handle unfamiliar tasks.
💡Real-World Example
A spam filter trained on thousands of email examples learns the general patterns of unwanted messages. When a brand-new type of scam email arrives that wasn't in any of the training data, the filter can still recognize it as spam because it understands the underlying tricks spammers use—without needing someone to manually show it examples of that specific scam first.
Related Terms
What did you think of our explanation?