I have been reading Penrose’s The Emperor’s New Mind, and about half way through he points out that quantum systems are linear until they are measured. From a machine learning standpoint this is not good. Linear artificial neural networks are of limited use and can only be used to model the simplest data sets. When a non-linear transform (activation function) is added after each linear layer in a neural network then it is able to solve difficult problems. To make quantum systems useful it may be necessary to use measurement as an activation function. An isolated quantum system would be allowed to evolve for some time and then be measured. The result would be fed into another quantum system. In this way the overall system would be layered like a neural network with measurement operations between the layers. Since quantum mechanics governs matter this means that most everything is a quantum neural network.