Cross-Domain Few-Shot Learning with Meta Fine-Tuning

TL;DR

This paper introduces a novel meta fine-tuning approach combining transfer learning and meta-learning techniques, significantly improving cross-domain few-shot learning performance on a challenging benchmark.

Contribution

It proposes a new method integrating transfer learning with meta-learning, using a modified episodic training with MAML and GNNs for better cross-domain adaptation.

Findings

Achieved 73.78% average accuracy on the benchmark.

Significant accuracy boost when combined with data augmentation.

Outperformed existing methods trained only on miniImagenet.

Abstract

In this paper, we tackle the new Cross-Domain Few-Shot Learning benchmark proposed by the CVPR 2020 Challenge. To this end, we build upon state-of-the-art methods in domain adaptation and few-shot learning to create a system that can be trained to perform both tasks. Inspired by the need to create models designed to be fine-tuned, we explore the integration of transfer-learning (fine-tuning) with meta-learning algorithms, to train a network that has specific layers that are designed to be adapted at a later fine-tuning stage. To do so, we modify the episodic training process to include a first-order MAML-based meta-learning algorithm, and use a Graph Neural Network model as the subsequent meta-learning module. We find that our proposed method helps to boost accuracy significantly, especially when combined with data augmentation. In our final results, we combine the novel method with the baseline method in a simple ensemble, and achieve an average accuracy of 73.78% on the benchmark. This is a 6.51% improvement over existing benchmarks that were trained solely on miniImagenet.