How Sensitive are Meta-Learners to Dataset Imbalance?

TL;DR

This paper investigates how dataset imbalance affects meta-learning for few-shot learning, revealing that meta-learners are surprisingly robust to dataset imbalance compared to task-level imbalance, highlighting their ability to generalize.

Contribution

It provides the first comprehensive analysis of dataset imbalance impact on meta-learning, demonstrating robustness of ML methods under various imbalance conditions.

Findings

Meta-learners are more robust to dataset imbalance than task imbalance.

Robustness persists even in long-tail and highly imbalanced datasets.

Meta-learning can learn generalizable features despite dataset imbalance.

Abstract

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 ($\rho<20$), with the effect holding even in long-tail datasets under a larger imbalance ($\rho=65$). 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.