Section outline

  • Highlighted
    • Le9 - Mon Nov. 7, 2022 Community detection problem: Granovetter's view and the role of weak ties; The core-periphery model; Network community profiles: V shape and nestedness in social networks; The overlapping community model; Agglomerative and divisive dendrograms; Recap of the dendrogram approach.


    • Le11 - Mon Nov. 14, 2022 Edge betweenness and the Girvan-Newman solution; Modularity as a measure for selecting the number of communities; Modularity in the two-communities case; Spectral approach and fast modularity optimisation; The Louvain algorithm; Main characteristics of the Louvain algorithm; Solving greediness by consensus clustering.

    • Le12 - Tue Nov. 15, 2022 Solving greediness by consensus clustering; The resolution limit; Generalized modularity; The role of the main parameter; Variation of information as a measure of dissimilarity; Using VI to optimally select the parameter; Example of application of the Louvain approach.

    • Le14 - Tue Nov. 22, 2022 The normalised cut criterion; Conductance. Suboptimal solution to normalised cut: algebraic connectivity and Fiedler's eigenvector of the normalised Laplacian matrix; The spectral clustering algorithm; Examples; Eigen-gap for eigenvector selection; Cheeger's inequality; Authority shift algorithms; Clique percolation; The BigClam: rationale of a model-based approach. 

    • Le16 - Tue Nov. 29, 2022 Non-interdisciplinary projects presentations: Iran revolts, political debate, foodwebs, stock markets, partner violence; The BigClam: algorithm and performance.

    • Le18 - Tue Dec. 6, 2022 The layout problem; Aesthetic criteria; Spring-embedder algorithm; Repulsive and attractive forces; Fruchterman & Reingold; Gravity; Force atlas 2; A comparison; BERT topic detection; UMAP as a force-layout algorithm.