Centerus is designing a deep neural net along some of the ideas in ART. One of the most important features of the Centerus net is the simplicity of its learning mechanism. It uses a slightly modified Hebbian learning mechanism, which allows for negative weights.