Few-Shot Learning with Graph Neural Networks

10 Nov 2017  ·  Victor Garcia, Joan Bruna ·

We propose to study the problem of few-shot learning with the prism of inference on a partially observed graphical model, constructed from a collection of input images whose label can be either observed or not. By assimilating generic message-passing inference algorithms with their neural-network counterparts, we define a graph neural network architecture that generalizes several of the recently proposed few-shot learning models. Besides providing improved numerical performance, our framework is easily extended to variants of few-shot learning, such as semi-supervised or active learning, demonstrating the ability of graph-based models to operate well on 'relational' tasks.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Cross-Domain Few-Shot ChestX GNN 5 shot 25.27 # 5
Cross-Domain Few-Shot EuroSAT GNN 5 shot 83.64 # 7
Cross-Domain Few-Shot ISIC2018 GNN 5 shot 43.94 # 9
Few-Shot Image Classification Stanford Cars 5-way (1-shot) GNN++ Accuracy 55.85 # 4
Few-Shot Image Classification Stanford Cars 5-way (5-shot) GNN++ Accuracy 71.25 # 4

Results from Other Papers


Task Dataset Model Metric Name Metric Value Rank Source Paper Compare
Few-Shot Image Classification Stanford Dogs 5-way (5-shot) GNN++ Accuracy 62.27 # 4

Methods


No methods listed for this paper. Add relevant methods here