Unlearning in feed-forward multi-nets

Multi-nets promise an improved performance over monolithic neural networks by virtue of their distributed implementation. Modular neural networks are multi-nets based on an judicious assembly of functionally different parts. This can be viewed as again a monolithic network, but with more complex neurons (the neural modules). Therefore they will share the same learning problems, notably the unlearning effect. In this paper we will look more closely into the reasons for unlearning and discuss how this can be applied to detect novelties.