1. Introduction to Graph Neural Networks (GNNs)
- Introduction to Message Passing with Self-Loops
- Generalized Neighborhood Aggregation
- Introduction to Graph Convolutional Networks (GCNs)
- Introduction to Graph Attention Networks (GATs)
- Heterogeneous Graphs and Graph Neural Networks
2. Fundamental Tasks in Graph Neural Networks
- Types of Supervised Tasks for Graph-Based Models
- Node and Graph Classification Using GNNS
- Graph Classification Techniques
- Model Architecture for Graph Classification
3. Challenges and Solutions in GNNs
- Over-Smoothing in GNN and the PairNorm Solution
- Challenges in Scaling Graph Neural Networks (GNNs) to Large Graphs
4. Advanced Techniques in GNNs
- Extensions of GraphSAGE Over Traditional GCNs
- FastGCN (Fast Graph Convolutional Networks) and Importance Sampling
- Sampling Paradigms in Graph Neural Networks for Large Graphs
- Cluster-GCN and Graph-Wise Sampling in GNN
- Spatial and Spectral Approaches to GNN
- Edge-Conditioned Convolution in GNNs
5. Training and Optimization in GNNs
6. Graph Generation Techniques
- Traditional Graph Generation Approaches
- Stochastic Block Models (SBMs) in Graph Generation
- Preferential Attachment Models (PA) in Graph Generation
- Deep Generative Models for Graphs
- Autoregressive Models for Graph Generation
- Generative Adversarial Networks (GANs) for Graphs
7. Variational Autoencoders (VAEs) and Graph-Based Generative Models
- Variational Autoencoders (VAEs) for Graph Generation
- Graph-Level Reconstruction in Graph-Based VAEs
- Node-Level Reconstruction in Graph-Based VAEs
- Reconstruction Loss in Graph-Based Variational Autoencoders (VAEs)
- Variational Inference in Deep Generative Models
- Latent Variables and Latent Space in Graph Models
- Binary Cross-Entropy (BCE) in Graph Generation