Welcome to the official repository for the Geometric Deep Learning course at the Moscow Institute of Physics and Technology (MIPT). This repository contains lecture materials, code examples, assignments, and projects related to learning deep learning methods on non-Euclidean domains such as graphs and manifolds.
Geometric Deep Learning extends traditional deep learning techniques to structured data like graphs, meshes, and manifolds. This course introduces the theoretical foundations, practical implementations, and applications of models that operate on these domains.
- Graph Neural Networks (GNNs)
- Spectral vs. Spatial methods
- Message Passing and Attention Mechanisms
- Graph Convolutions
- Node Classification & Graph Classification
- Embedding Techniques (Node2Vec, GCN, GAT, etc.)
- Equivariance and Invariance
- Applications in:
- 3D Computer Vision
- Molecular Chemistry & Drug Discovery
- Physical Simulations
- Python
- PyTorch
- PyTorch Geometric (PyG)
- NetworkX
- NumPy / SciPy
- Matplotlib & Seaborn