How Sensitive are Meta-Learners to Dataset Imbalance?
Not on the map. Why?
Abstract (arXiv)
Meta-Learning (ML) has proven to be a useful tool for training Few-Shot Learning (FSL) algorithms by exposure to batches of tasks sampled from a meta-dataset. However, the standard training procedure overlooks the dynamic nature of the real-world where object classes are likely to occur at different frequencies. While it is generally understood that imbalanced tasks harm the performance of supervised methods, there is no significant research examining the impact of imbalanced meta-datasets on the FSL evaluation task. This study exposes the magnitude and extent of this problem. Our results show that ML methods are more robust against meta-dataset imbalance than imbalance at the task-level with a similar imbalance ratio (ρ<20), with the effect holding even in long-tail datasets under a larger imbalance (ρ=65). Overall, these results highlight an implicit strength of ML algorithms, capable of learning generalizable features under dataset imbalance and domain-shift. The code to reproduce the experiments is released under an open-source license.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 22 of its 38 references that are in the map. First posted on arXiv 12 Apr 2021.
Builds on
22 of 38 references.
References · 22
- 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
- 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 7,005
- 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,897
- 2017Prototypical Networks for Few-shot Learningcited by 3,015
- 2016Matching Networks for One Shot Learningcited by 2,255
- 2017Learning to Compare: Relation Network for Few-Shot Learningcited by 1,143
- 2020Meta-Learning in Neural Networks: A Surveycited by 950
- 2019Large-Scale Long-Tailed Recognition in an Open Worldcited by 705
- 2019A Closer Look at Few-shot Classificationcited by 684
- 2017A systematic study of the class imbalance problem in convolutional neural networkscited by 513
- 2019Meta-Dataset: A Dataset of Datasets for Learning to Learn from Few Examplescited by 315
- 2019A Baseline for Few-Shot Image Classificationcited by 298
10 more
- 2019SimpleShot: Revisiting Nearest-Neighbor Classification for Few-Shot Learningcited by 195
- 2016Towards a Neural Statisticiancited by 121
- 2019Learning to Balance: Bayesian Meta-Learning for Imbalanced and Out-of-distribution Taskscited by 48
- 2021Shallow Bayesian Meta Learning for Real-World Few-Shot Recognitioncited by 48
- 2020Bayesian Few-Shot Classification with One-vs-Each P'olya-Gamma Augmented Gaussian Processescited by 32
- 2020Learning to Stop While Learning to Predictcited by 31
- 2021ORBIT: A Real-World Few-Shot Dataset for Teachable Object Recognitioncited by 29
- 2020Defining Benchmarks for Continual Few-Shot Learningcited by 21
- 2020Contextualizing Enhances Gradient Based Meta Learningcited by 3
- 2020A Comparison of Few-Shot Learning Methods for Underwater Optical and Sonar Image Classificationcited by 2
Built on by
Citing papers per year: 2022: 1.
1 paper.