DAMSL: Domain Agnostic Meta Score-based Learning

TL;DR

DAMSL introduces a domain-agnostic meta-learning approach that leverages score-based representations and graph neural networks to improve cross-domain few-shot learning performance significantly.

Contribution

It proposes a novel domain-agnostic metric space and GNN-based embedding method to address overfitting and under-utilization issues in prior meta-learning techniques.

Findings

Outperforms state-of-the-art methods on established benchmarks

Achieves significant accuracy improvements across various domain shifts

Effectively utilizes support set structure for better generalization

Abstract

In this paper, we propose Domain Agnostic Meta Score-based Learning (DAMSL), a novel, versatile and highly effective solution that delivers significant out-performance over state-of-the-art methods for cross-domain few-shot learning. We identify key problems in previous meta-learning methods over-fitting to the source domain, and previous transfer-learning methods under-utilizing the structure of the support set. The core idea behind our method is that instead of directly using the scores from a fine-tuned feature encoder, we use these scores to create input coordinates for a domain agnostic metric space. A graph neural network is applied to learn an embedding and relation function over these coordinates to process all information contained in the score distribution of the support set. We test our model on both established CD-FSL benchmarks and new domains and show that our method overcomes the limitations of previous meta-learning and transfer-learning methods to deliver substantial improvements in accuracy across both smaller and larger domain shifts.