Multi-layer quantum neural network controller trained by real-coded genetic algorithm

We investigate a quantum neural network and discuss its application to controlling systems. First, we consider a multi-layer quantum neural network that uses qubit neurons as its information processing unit. Next, we propose a direct neural network controller using the multi-layer quantum neural network. To improve learning performance, instead of applying a back-propagation algorithm for the supervised training of the multi-layer quantum neural network, we apply a real-coded genetic algorithm. To evaluate the capabilities of the direct quantum neural network controller, we conduct computational experiments controlling a discrete-time nonlinear system and a nonholonomic system (a two-wheeled robot). Experimental results confirm the effectiveness of the real-coded genetic algorithm in training a quantum neural network and prove the feasibility and robustness of the direct quantum neural network controller.