# A Comprehensive Survey of Few-shot Learning: Evolution, Applications, Challenges, and Opportunities

**Authors:** Yisheng Song, Ting Wang, Subrota K Mondal, Jyoti Prakash Sahoo  
**arXiv:** [2205.06743](https://arxiv.org/abs/2205.06743)  
**Date:** 2022-05-13  
**Summary page:** https://www.summarizepaper.com/en/paper/2205.06743v1/

## Key points

- Few-shot learning (FSL) is an effective learning method with great potential
- The authors investigate over 200 latest papers on FSL to provide an overview of recent advances and comparisons of existing works
- Elaboration and comparison of similar concepts including few-shot learning, transfer learning, and meta-learning
- Proposal of a novel taxonomy to classify existing work based on the level of abstraction of knowledge in accordance with the challenges of FSL
- In-depth analysis and discussion about recent advances in cross-domain learning, fine-tuning, meta-learning, and transfer learning
- Highlighting important applications of FSL in computer vision
- Coverage of various research hotspots within the field
- Unique insights into technology evolution trends and potential future research opportunities
- Aim to provide guidance for follow-up research in FSL by presenting a timely and comprehensive overview along with unbiased comparisons
- Beneficial for researchers interested in exploring few-shot learning methods as it provides an extensive overview that can help gain insight into current trends and identify potential areas for further study.

## Summary

Few-shot learning (FSL) has emerged as an effective learning method with great potential. In this comprehensive survey, the authors investigate over 200 latest papers on FSL published in the past three years to provide an overview of recent advances and impartial comparisons of existing works. They first elaborate and compare similar concepts including few-shot learning, transfer learning, and meta-learning to avoid conceptual confusion. The authors also propose a novel taxonomy to classify existing work based on the level of abstraction of knowledge in accordance with the challenges of FSL. The survey provides in-depth analysis and insightful discussion about recent advances in each subsection, enriching the understanding of topics such as cross-domain learning, fine-tuning, meta-learning, and transfer learning. The authors highlight the important application of FSL in computer vision and cover various research hotspots within this field. The survey concludes with unique insights into technology evolution trends and potential future research opportunities. It aims to provide guidance for follow-up research in FSL by presenting a timely and comprehensive overview of recent advances along with unbiased comparisons of strengths and weaknesses. Overall, this survey contributes to the understanding of FSL by providing a detailed analysis of recent advancements, highlighting important applications, and identifying future research directions. This survey is beneficial for researchers who are interested in exploring few-shot learning methods as it provides an extensive overview that can help them gain insight into current trends and identify potential areas for further study.