Attentive Graph Neural Networks for Few-Shot Learning

This repository contains the code for Attentive Graph Neural Networks for Few-Shot Learning.

Running the code

Preliminaries

Environment

Python 3.7.3
Pytorch 1.7.1
tensorboardX

Datasets

miniImageNet: the Google Drive or Baidu Drive (uk3o) for downloading (courtesy of DeepEMD)
tieredImageNet (courtesy of Kwonjoon Lee)

Download the datasets and link the folders into materials/ with names mini-imagenet, tiered-imagenet and imagenet. Note imagenet refers to ILSVRC-2012 1K dataset with two directories train and val with class folders.

When running python programs, use --gpu to specify the GPUs for running the code (e.g. --gpu 0,1). For Classifier-Baseline, we train with 4 GPUs on miniImageNet and tieredImageNet and with 8 GPUs on ImageNet-800. Meta-Baseline uses half of the GPUs correspondingly.

In following we take miniImageNet as an example. For other datasets, replace mini with tiered. By default it is 1-shot, modify shot in config file for other shots. Models are saved in save/.

*The models on miniImageNet and tieredImageNet use ConvNet-4 as backbone, the channels in each block are 64-96-128-256.

1. Pretraining the ConvNet4 Backbone

python train_classifier.py --config configs/train_classifier_mini.yaml

The pretrained Classifier-Baselines can be downloaded from Google Drive (Mini-ImageNet , Tiered-ImageNet).

You can unzip and place the foder under the 'save' folder.

2. Training and Testing AGNN

python train_meta.py --config configs/train_meta_mini.yaml

Citation

@inproceedings{cheng2022attentive,
  title={Attentive graph neural networks for few-shot learning},
  author={Cheng, Hao and Zhou, Joey Tianyi and Tay, Wee Peng and Wen, Bihan},
  booktitle={2022 IEEE 5th International Conference on Multimedia Information Processing and Retrieval (MIPR)},
  pages={152--157},
  year={2022},
  organization={IEEE}
}

Extension is accepted by IEEE Transactions on Multimedia, Update Soon

@article{cheng2023graph,
  title={Graph Neural Networks With Triple Attention for Few-Shot Learning},
  author={Cheng, Hao and Zhou, Joey Tianyi and Tay, Wee Peng and Wen, Bihan},
  journal={IEEE Transactions on Multimedia},
  year={2023},
  publisher={IEEE}
}

Acknowledgment

We thank the following repos providing helpful components/functions in our work.

chenghao-ch94/AGNN