Few-Shot Learning

Quick Definition

Learning to perform tasks from a small number of examples

A ML paradigm where models learn to perform tasks from a small number of examples.

in-context learning, one-shot classification

in-context-learning transfer-learning

Mechanism allowing sequence positions to attend to all others

Neural network for image and grid-like data processing

Learned vector representations of discrete objects in continuous space

Research ensuring AI is developed safely without harm

Architecture using self-attention for parallel processing

Autonomous software taking goal-directed actions