In this project I made at EURECOM university, in collaboration with the research department and a major player in the IT and data market, I have carried out:
- A deep theoretical analysis of Graph Convolutional Networks
- The reproduction of the results presented in the milestone paper by Kipf and Welling "Semi-Supervised Classification with Graph Convolutional Networks" (ICLR), https://arxiv.org/abs/1609.02907.
- A research on the pitfalls of anomaly detection on huge graphs (like financial graphs)
- The implementation of a solution of non-trivial problem of minibatching during GCN training, following the intuition of Frasca et al. "SIGN: Scalable Inception Graph Neural Networks", https://arxiv.org/abs/2004.11198.
In this repository you can also find my report, concerning the whole analysis and research I carried out (GCN__AnomDetect_EURECOM).