Comparative Analysis of Gradient Methods for Source Identification in a Diffusion-Logistic Model

Mathematical modeling of information diffusion in Twitter

“…-particle swarm optimization (PSO) is a stochastic method for finding the global minimum region; -multilevel gradient method (MGM) [4] is a modification of the local gradient descent method which has a higher rate of convergence; -tensor train (TT) [5] is a global optimization method based on tensor decomposition; -combination of TT and MGM as a hybrid optimization approach. Results: For the benchmark problem the source (solid black line) of information propagation was specified from additional information about the process at 21 points in time for each hour.…”