FSL-Survey[1] is a survey on Few-Shot Learning(FSL), which cotains 166 paper to review Few-Shot Learning. They categorize FSL methods from three perspectives: data, model and algorithm. There are some details of reading it.

Contents

• Paper & note
• Paper understanding
• Note
• References

Paper & note

Paper: Generalizing from a Few Examples: A Survey on Few-Shot Learning(CSUR 2019 paper)
Note: Mendeley

Paper

Abstract

1. Starting from a formal definition of FSL, then point out that the core issue in FSL.
2. Data: which uses prior knowledge to augment the supervised experience.
3. model: which uses prior knowledge to reduce the size of the hypothesis space.
4. algorithm: which uses prior knowledge to alter the search for the best hypothesis in the given hypothesis space.
5. Promising directions are also proposed to provide insights for future research.

Definition

Taxonomy

Data

Model

Multitask Learning

Embedding Learning

Learning with External Memory

Generative Modeling

Algorithm

Refining Existing Parameters

Refining Meta-Learned Parameter

Learning the Optimizer

Meta-learning

Note

Offical Online link can be found in FewShotPapers[2].

References

[1] Wang Y, Yao Q, Kwok J T, et al. Generalizing from a few examples: A survey on few-shot learning[J]. ACM Computing Surveys (CSUR), 2019.
[2] FewShotPapers. https://github.com/tata1661/FewShotPapers