graph-classification

Graph embeddings (Graph2Vec, DeepWalk, NetLSD) vs. GNNs on TU datasets. Team project for Information Systems, ECE NTUA.

Graph Pooling Lab (GPLab) is a benchmark framework for evaluating hierarchical graph pooling methods under controlled and comparable experimental settings.

graphotaxy performs undirected graph classification.

Semi-supervised Label Propagation Algorithm (LPA) for graph node classification and community clustering

Semi-supervised Label Propagation Algorithm (LPA) for graph node classification and community clustering

GCN, GAT, GIN, and GraphSAGE graph-classification benchmarks with upstream multi-seed results.

A scikit-learn compatible library for graph kernels

GraphTreeBoost is a gradient-boosted graph learning framework combining soft decision trees with spectral feature aggregation. Chebyshev and heat-kernel filters enable efficient graph-aware learning without eigendecomposition. Experiments on six benchmarks demonstrate improvements over feature-only baselines.

Persistent-JEPA: a persistent graph world model for continual developmental learning.

Operational Runtime Behavior Mining for Open-Source Supply Chain Security

Comparative study of graph classification methods on MUTAG using graph kernels, embeddings, and GNNs.

OpenGraphXAI collection of benchmarks for XAI in graph classification

Data preprocessing and training of a Deep Learning graph architecture for the classification of tumor types, with focus on the most impactful genomic and clinical differences through model explainability. Use cases tested: Lung cancer (LUAD / LUSC) and Kidney cancer (KIRC / KICH / KIRP). Project for the 'AI for Bioinformatics' course.

DRESS: A Continuous Framework for Structural Graph Refinement

Graph Perceiver IO paper code : graph multimodal learning, multimodal learning with graph

Pattern Mining for the Classification of Public Procurement Fraud

Implementation of the BNPool layer and code to reproduce the experiments in "Bayesian Nonparametric GNNs for graph pooling and clustering".

Official implementation of "Differentiable Lifting for Topological Neural Networks." A general framework (∂lift) for end-to-end learning of graph liftings to hypergraphs, cellular complexes, and simplicial complexes.

A minimal hybrid Quantum–Graph Neural Network prototype

Tiny HTTP framework built on node:http