This repository contains my coursework and projects for the NI-GNN (Graph Neural Networks) course starting 01.02.2025 to 23.05.2025.
Student Name: I Wayan Iswara Jay Junior
CTU email: [email protected]
TUM email: [email protected]
Student status: EuroTeQ course participant from TUM
- Topic: Exploration of the ICIJ Offshore Leaks dataset
- Dataset: ICIJ Offshore Leaks dataset
- Topic: GNN models comparison using dgl library
- Dataset: COX2 dataset from TUdataset (Binary graph classification)
- Architecture: GCN, GraphSAGE
- Highlights: Architecture comparison, training curves, confusion matrix
- Topic: Comparing explainer results for GNN models using pytorch-geometric library
- Dataset: Amazon Computers (Multiclass node classification)
- Methods: GNNExplainer
- Highlights: Feature importance, explanation subgraph, explanation fidelity, unfaithfulness analysis
- Topic: Multiclass graph classification for enzyme classification
- Dataset: ENZYMES from TUdataset
- Models: GCN, GraphSAGE, GIN, GATv2
- Results: Accuracy, explanation quality, confusion matrix
Should be given in each of homeworks/final project notebooks