eleGAN23 / GINN

Graph Imputation Neural Network

Graph Imputation Neural Networks (GINN)

This is the companion code for the paper: Missing Data Imputation with Adversarially-trained Graph Convolutional Networks, arXiv:1905.01907, 2019.

Imputing missing data with graph neural networks

We perform imputation of missing data in a generic dataset by (a) building a graph of similarities between examples, and (b) running an autoencoder with graph convolutions [1] on top of that.

Organization of the code

All the code for the models described in the paper can be found in ginn/core.py and ginn/models.py. Examples of use with accompanying notebooks are in examples.

References

[1] Kipf, T.N. and Welling, M., 2016. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907.