In 2022 U. Ernst, M. Keßeböhmer and M. Schünemann were able to prove a topological conjugation of a neural network, called the EHE model, to translation actions on the torus. This conjugation enabled a detailed investigation of characterizing properties of the neural network such as avalanche and assembly distributions.

In this talk we will generalize this topological conjugation to a broader set of neural networks, leading to possible applications for the Abelian sandpile model.