You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Multi-metric framework for quantitatively validating GNN explanations in brain tumor segmentation. It evaluates GraphSAGE, GAT, and ChebNet with Integrated Gradients, GraphLIME, and GNNExplainer on BraTS 2023 supervoxel graphs.
Project to understand and built a Graph Neural Network for molecular property prediction (toxicity, solubility, and drug-likeness) on MoleculeNet benchmarks.
Graph Neural Network fraud detection on the Elliptic Bitcoin dataset - GraphSAGE node classification vs. a flat XGBoost baseline, GNNExplainer explanations, FastAPI serving and a Streamlit subgraph dashboard. 100% free stack: PyTorch CPU + PyTorch Geometric, no paid APIs or compute.
Official implementation of Explainable Molecular Toxicity Prediction using Graph Neural Networks, GNNExplainer, PGExplainer, and attention-based message passing on the Tox21 benchmark.