Powering Finetuning in Few-Shot Learning: Domain-Agnostic Bias Reduction with Selected Sampling

TL;DR

This paper introduces a domain-agnostic bias reduction method for few-shot learning that combines a distribution calibration module and selected sampling to improve finetuning of deep networks, achieving state-of-the-art results across multiple datasets.

Contribution

It proposes a novel bias reduction framework using DCM and SS to enhance finetuning in few-shot learning, addressing domain and class-specific biases.

Findings

Achieves state-of-the-art results on Meta-Dataset across ten diverse datasets.

Effectively reduces domain shift and class bias during finetuning.

Demonstrates practical applicability with consistent performance improvements.

Abstract

In recent works, utilizing a deep network trained on meta-training set serves as a strong baseline in few-shot learning. In this paper, we move forward to refine novel-class features by finetuning a trained deep network. Finetuning is designed to focus on reducing biases in novel-class feature distributions, which we define as two aspects: class-agnostic and class-specific biases. Class-agnostic bias is defined as the distribution shifting introduced by domain difference, which we propose Distribution Calibration Module(DCM) to reduce. DCM owes good property of eliminating domain difference and fast feature adaptation during optimization. Class-specific bias is defined as the biased estimation using a few samples in novel classes, which we propose Selected Sampling(SS) to reduce. Without inferring the actual class distribution, SS is designed by running sampling using proposal distributions around support-set samples. By powering finetuning with DCM and SS, we achieve state-of-the-art results on Meta-Dataset with consistent performance boosts over ten datasets from different domains. We believe our simple yet effective method demonstrates its possibility to be applied on practical few-shot applications.