多尺度动态图信号分析matlab工具箱.zip

Multiscale Dynamic Graph Signal Analysis Toolbox (MDGABox) The official Graph Signal Processing Toolbox (GSPBox) needs to be initially installed on your computer for running the scripts in MDGABox. The GSPBox can be downloaded through the following link: https://epfl-lts2.github.io/gspbox-html/ ================================== After running the GSPBox on your computer, you can easily run and test the MDGABox toolbox as follows: 1- Run_MDGA_toolbox.m: This script is used to test and run the MDGABox toolbox. It includes two synthetic signals to illustrate the algorithm's performances. 2- MDGA.m: This script includes two approaches (i.e., the two-stage approach and DGMD algorithms) for Multiscale Dynamic Graph Mode Decomposition. 3- two_stage.m: This script solves two different optimization problems separately, that is SS-MVMD for mode decomposition and a graph learning algorithm (based on the dual-primal approach) for dynamic graph connectivity analysis. The two-stage approach firstly decomposes the dynamic graph signal into multiple oscillatory components, and then it uses those components for the multiscale analysis of the dynamic connectivity structures. 4- ssmvmd.m: The Simplified Successive Multivariate Variational Mode Decomposition (SS-MVMD) decomposes the dynamic graph signal into its multiple oscillatory components. 5- DGMD.m: This script solves two different optimization problems jointly, i.e., its cost function combines the constraints of the optimization formulation of the mode decomposition and the constraints of the optimization formulation of the graph learning algorithm (based on the dual-primal approach) for dynamic graph connectivity analysis. The DGMD method decomposes the dynamic graph signal into multiple oscillatory components and obtains their associated dynamic connectivity structures altogether. 6- learning_graph.m: This function computes a weighted adjacency matrix W from squared pairwise distances in Z of previous, current, and next windows of the signal, using the smoothness assumption. This code has been developed through modification in the 'GSP_LEARN_GRAPH_LOG_DEGREES.m' file obtained from the GSPBox. 7- data2psi.m: This function calculates the phase slope index (PSI). The PSI measure is used for calculating the adjacency (connectivity) matrices of the dynamic graph signals (multivariate signals). The PSI function has been developed and published by the researchers below: Nolte G, Ziehe A, Nikulin VV, Schl"ogl A, Kr"amer N, Brismar T, M"uller KR. Robustly estimating the flow direction of information in complex physical systems. 8- twostage_pli.m: This function calculates the phase lag index (PLI and wPLI). The PLI measure is used for calculating the adjacency (connectivity) matrices of the dynamic graph signals (multivariate signals). The PLI and wPLI measures were introduced by researchers in: [1] https://doi.org/10.1002%2Fhbm.20346 [2] https://doi.org/10.1016/j.neuroimage.2011.01.055 ================================== You can find the complete public MDGABox on mathworks.com: https://se.mathworks.com/matlabcentral/profile/authors/16833607 Author: Mojtaba Nazari and Naveed ur Rehman December 2023