Clusters and Communities

This lecture covers the concepts of clustering and community detection. Clustering involves finding sets of points close to each other in a distance metric, while community detection focuses on identifying highly interconnected nodes in a network. The K-means algorithm is introduced for clustering, along with its iterative approximation process. The lecture then transitions to Gaussian Mixture Models (GMM) for clustering, explaining the EM algorithm for GMM. Modularity is discussed as a measure of community strength, with an emphasis on interpreting and calculating modularity. The Louvain method for community detection is presented as a bottom-up approach to building a hierarchy of communities. The lecture concludes by highlighting the challenges of unsupervised techniques and providing references for further study.