Zero-Shot Learning
Quick Definition
Performing tasks on unseen classes without training examples
Full Definition
A ML paradigm where a model performs tasks on unseen classes without any training examples for those classes.
Examples
image classification of new categories, cross-domain transfer
Related Terms
few-shot-learning
transfer-learning
meta-learning
More Artificial Intelligence Terms
Prompt Injection
Attack manipulating AI behavior by overriding instructions
Reward Modeling
Training models to predict human preferences for RLHF
Recurrent Neural Network
Neural network with cyclic connections for sequence processing
Hallucination
AI generating confident but factually incorrect information
Adversarial Attack
Crafting inputs that fool AI models into wrong predictions
Neural Architecture Search
Automating the design of optimal neural network architectures