Course II - Keith Levin/Jesús Arroyo/Joshua Cape - Statistical Network Analysis

Statistical Network analysis
Lecturers: Keith Levin/Jesús Arroyo/Joshua Cape (TexasA&M/Wisconsin Madison)

This  course provides a more in-depth, detailed view of statistical problems that arise for network-valued data and random graph models. Topics include parameter estimation, hypothesis testing, inference, error bounds for clustering, information theoretic limits, and algorithms for computation. The first set of three lectures will focus on models and methods for network data analysis, with an emphasis on two classical problems: graph matching and community detection. The second set of three lectures will focus on spectral methods for network analysis, namely techniques rooted in matrix decompositions, perturbations, and factorizations, with an emphasis on spectral clustering and embedding.

Lecture 1 (Keith Levin) Introduction to Statistical Methods. Stochastic Block Model. Semidefinite Programing. Belief Propagation. Approximate Message Passing. Spectral Methods.
Lecture 2 (Keith Levin)
Lecture 3  (Keith Levin)
Lecture 4  (Keith Levin)
Lecture 5 (Jesús Arroyo) Statistical Network analysis Graph Matching
Lecture 6 (Jesús Arroyo) Statistical Network analysis Community Detection I
Lecture 7 (Jesús Arroyo) Statistical Network analysis Community Detection II
Lecture 8: (Joshua Cape) Spectral Clustering with Graph Laplacians
Lecture 9: (Joshua Cape) Spectral Clustering from adjacency Matrices
Lecture 10: (Joshua Cape) Perturbation