Various Network Science Projects (2021-2022)
This repo demonstrates multiple Network Science Techniques including:
- introduction to network science, Graph manipulation, extracting general statistics about graphs and graph visualization.
- Power law degree functions and degree distributions.
- Random Graphs generation including Erdos-Renyi model, hyperparameter optimization for the random graph, and general statistics of random graphs.
- More Random Graphs models including Barabasi-Albert model and Watts-Strogatz model with their respective statistics.
- A deep dive into Centrality measures and Page-Rank Algorithm on the Moscow metro station dataset.
- Study structural similarities on co-watch dataset such as Adjacency matrix, Person correlation, Jaccard and Cosine similarity and Assortative Mixing.
- Graph partitioning algorithms including Newman, Modularity, Eigenvalues Algorithms and Spectral clustering.
- Community detection Algorithms including Agglomerative, Louvian, Ego-Splitting and Label propagation methods.
- Simulation of Compartmental Epidemic including Euler, SI, SIS, SIR and SIRS models.
- Applying SIS and SIRS models on Graphs. Implementing Random and Selective immunization.
- Influence Propagation effects on Graphs including the study of Linear threshold and independent cascade models and influence maximization problem.
- Applying ML on Graphs including Node Classification, Label Propagation by random_walk, SVD node embeddings and the Deepwalk model.
- Link Prediction on Email network dataset using similarity score, Dot product predictor and edges embeddings.
- Deep dive into graph embeddings using DeepWalk, Node2Vec, Hierarchical softmax and GraRep methods.
- A Study on the different types of GNN (Graph neural networks) on the Cora dataset.
- Knowledge Graphs models including: Translation models, Entitiy embeddings, Nearest neighbors and Tail predictions.
Skills developed: pandas | scikit-learn | matplotlib | numpy | Network Science | Node Classifications | GNN | pytorch | networkx | python | Graph and Node embeddings | Graph Statistics | Knowledge Graphs.
This repo is part of the NS course, HSE, Moscow, Russia.