The code in this repository implements a metric learning approach for irregular graphs. The method has been applied on brain connectivity networks and is presented in our papers:
-
Sofia Ira Ktena, Sarah Parisot, Enzo Ferrante, Martin Rajchl, Matthew Lee, Ben Glocker, Daniel Rueckert, Metric learning with spectral graph convolutions on brain connectivity networks, NeuroImage, 2018.
-
Sofia Ira Ktena, Sarah Parisot, Enzo Ferrante, Martin Rajchl, Matthew Lee, Ben Glocker, Daniel Rueckert, Distance Metric Learning using Graph Convolutional Networks: Application to Functional Brain Networks, Medical Image Computing and Computer-Assisted Interventions (MICCAI), 2017.
The code is released under the terms of the MIT license. Please cite the above paper if you use it.
There is also implementations of the filters and graph coarsening used in:
- Michaël Defferrard, Xavier Bresson, Pierre Vandergheynst, Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering, Neural Information Processing Systems (NIPS), 2016.
The implementaton of the global loss function is based on:
- Vijay Kuma, Gustavo Carneiro, Ian Reid, Learning Local Image Descriptors with Deep Siamese and Triplet Convolutional Networks by Minimising Global Loss Functions, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
-
Clone this repository.
git clone https://github.com/sk1712/gcn_metric_learning cd gcn_metric_learning
-
Install the dependencies. Please edit
requirements.txt
to choose the TensorFlow version (CPU / GPU, Linux / Mac) you want to install, or install it beforehand.pip install -r requirements.txt # or make install
To use our siamese graph ConvNet on your data, you need:
- pairs of graphs as matrices where each row is a node and each column is a node feature,
- a class label for each graph,
- an adjacency matrix which provides the structure as a graph; the same structure will be used for all samples.
Please get in touch if you are unsure about applying the model to a different setting.