Few-Shot Learning (FSL)

What is Few-Shot Learning (FSL)?

Few-shot Learning (FSL) is an approach where the training dataset contains a handful or a ‘few’ annotated examples. In general, its common practice to feed machine learning models as much data as the model can take. Few-shot learning takes a different approach and aims to build accurate models with much less data by focusing on learning more specific representations of the data for the task at hand.

In Legal AI and Legal NLP, Few-shot Learning is of particular importance since it can greatly reduce the need for labelled training data. With a low bar to entry, Few-shot Learning means that models can be trained with only a handful of labelled examples

Related Terms