Stable Growing Neural Gas

I got a flash of interest when looking at algorithms that support learning the topology of a 3D world by imitating a teacher without having any knowledge about it before starting. Making these kind of algorithms well suited for learning topologies as they go. The algorithm Stable Growing Neural Gas is an alteration of Growing Neural Gas and was proposed here. In the prototype like implementation done by me, you can see how the gas grows to cover the circle in which the teacher is moving.

Implementation details:
Language: C++ with help of SFMLAnt Tweak Bar and Igloo

 

lYwBE8B