Fig. 1: Flowchart of the studied embedding algorithms.
From: Model-independent embedding of directed networks into Euclidean and hyperbolic spaces
The left side of the figure traces the algorithmic steps for creating a hyperbolic embedding with the High-Order Proximity preserved Embedding (HOPE), our TRansformation of EXponential shortest Path lengths to EuclideaN measures (TREXPEN) and their variants by converting the Euclidean node arrangement obtained from them to a hyperbolic one with our model-independent conversion (MIC). The right side of the figure shows the algorithmic steps of our method named TRansformation of EXponential shortest Path lengths to hyperbolIC measures (TREXPIC), which embeds networks directly in the hyperbolic space. The embedding parameters are written in red: the parameters α and q adjust how the elements of the reduced matrices depend on the distances measured along the graph to be embedded, d denotes the number of dimensions of the embedding space, ζ (usually set to 1) tunes the curvature of the hyperbolic space, and C (usually set to 2) controls the extent of the graph in the hyperbolic space when using MIC.
