Generalizing from a Few Examples: A Survey on Few-Shot Learning

By Y. Wang et al

Published on March 10, 2020

Read the original document by opening this link in a new tab.

Table of Contents

1. Introduction
2. Overview
2.1 Problem Definition
2.2 Relevant Learning Problems
3. Generalizing to New Tasks
4. Core Issue in FSL
5. Categorization of FSL Methods
6. Future Directions
7. Conclusion

Summary

Machine learning has been successful in data-intensive applications, but faces challenges with small datasets. Few-Shot Learning (FSL) addresses this by rapidly generalizing to new tasks with limited samples. The paper provides a thorough survey of FSL, defining it and categorizing methods based on data, model, and algorithm perspectives. It discusses the core issue of unreliable risk minimizer in FSL and proposes future research directions.