2.0. Foundations of Classical Graph Signal Processing: Mapping a discrete-time signal onto a graph, signal shifting over a graph, from one-to-one mappings to one-to-many.
2.1. Graph Fourier Transform (GFT): From DFT to GFT and back to DFT as a special case, graph spectrum, inverse transform (IGFT).
2.2. Definition of Systems on Graphs: Linear shift-invariant systems, convolution, graph Z-transform, system transfer function, filter design, and signal filtering on graphs.
Statistical Graph Signal Processing: Basic concepts of stochastic processes, definition of stationarity based on the adjacency matrix and Laplacian, definition of wide-sense stationary graphs based on shifting in the spectral domain, Wiener filters on graphs, periodogram, moving average models, autoregressive models, autoregressive moving average models.
Sampling in the Node Domain. Sampling in the Frequency Domain of the Graph. Graph Reconstruction.
Graph Development from Data: Approaches based on real-world models. Presentation of statistical approaches (Markov Random Fields, Bayesian networks). Models based on continuity constraints.
Convolutional Neural Networks using Graphs: Activation functions, multi-layer networks.
Applications: Sensor placement, matrix completion, sampling of 3D point clouds, image, and time series representation.