Edge-labeling Graph Neural Network for Few-shot Learning

TL;DR

This paper introduces an edge-labeling graph neural network (EGNN) for few-shot learning that explicitly models class clusters by predicting edge labels, improving performance on image classification tasks.

Contribution

The paper presents a novel EGNN that predicts edge labels for explicit clustering, enabling better few-shot learning without retraining for different class numbers.

Findings

EGNN outperforms existing GNNs on benchmark datasets.

It effectively models intra-cluster similarity and inter-cluster dissimilarity.

EGNN is adaptable to various class numbers without retraining.

Abstract

In this paper, we propose a novel edge-labeling graph neural network (EGNN), which adapts a deep neural network on the edge-labeling graph, for few-shot learning. The previous graph neural network (GNN) approaches in few-shot learning have been based on the node-labeling framework, which implicitly models the intra-cluster similarity and the inter-cluster dissimilarity. In contrast, the proposed EGNN learns to predict the edge-labels rather than the node-labels on the graph that enables the evolution of an explicit clustering by iteratively updating the edge-labels with direct exploitation of both intra-cluster similarity and the inter-cluster dissimilarity. It is also well suited for performing on various numbers of classes without retraining, and can be easily extended to perform a transductive inference. The parameters of the EGNN are learned by episodic training with an edge-labeling loss to obtain a well-generalizable model for unseen low-data problem. On both of the supervised and semi-supervised few-shot image classification tasks with two benchmark datasets, the proposed EGNN significantly improves the performances over the existing GNNs.