Active Few-Shot Learning with FASL

TL;DR

This paper introduces FASL, a platform combining active and few-shot learning for efficient text classification with minimal annotations, including methods to select data and determine stopping points.

Contribution

It presents a novel platform that integrates active and few-shot learning for rapid text classification model development, with a new stopping prediction model.

Findings

Active learning methods vary in effectiveness in few-shot settings.

FASL enables faster model training with fewer annotations.

A model for predicting when to stop annotation improves efficiency.

Abstract

Recent advances in natural language processing (NLP) have led to strong text classification models for many tasks. However, still often thousands of examples are needed to train models with good quality. This makes it challenging to quickly develop and deploy new models for real world problems and business needs. Few-shot learning and active learning are two lines of research, aimed at tackling this problem. In this work, we combine both lines into FASL, a platform that allows training text classification models using an iterative and fast process. We investigate which active learning methods work best in our few-shot setup. Additionally, we develop a model to predict when to stop annotating. This is relevant as in a few-shot setup we do not have access to a large validation set.