License : Creative Commons Attribution 4.0 International (CC BY-NC-SA 4.0)
Copyright : Hervé Frezza-Buet, CentraleSupelec
Last modified : November 16, 2022 11:49
Link to the source : gngt.md

Growing Neural Gas with Targetting (GNG-T)

The target controls of the number of prototypes, independently from the sampling density. We label the connected components (with colors).

As the same structure is modified from one frame to the next, we can keep trace of the speeds (gathered by connected components as previously). We use a Savitsky-Golay filter to remove the hustle and bustle of the node positions (high frequence in the positions) in order to compute speeds.

Hervé Frezza-Buet,