Permutation Entropy for Graph Signals

Entropy metrics (for example, permutation entropy) are nonlinear measures of irregularity in time series (one-dimensional data). Some these entropy can be generalised to data on periodic structures such as a grid or lattice pattern (two-dimensional data) using its symmetry, thus enabling their application images. However, have not been developed for signals sampled irregular domains, defined by graph. Here, we define the first an metric analyse measured over graphs generalising entropy, well-established based comparison neighbouring values within patterns series. Our algorithm is comparing signal nodes, adjacency matrix. We show that this generalisation preserves properties classical and recent images, it applied any graph structure with synthetic real signals. expect present work enable extension other dynamic approaches