Proposal and Evaluation of A Neural Network by Qubit Neuron.

Noriaki Kouda, Nobuyuki Matsui, H. Nishimura

We have proposed a novel neuron model whose states and interactions with other neurons are based on the laws of quantum physics. In our previous works, neural network constricted by this neuron, i.e., qubit neural network, has been shown through various benchmark simulations. As an example of nonholonomic system control, in this report, the swing-up control of inverted pendulum by the qubit neural network is investigated to estimate the efficiencies for more practical applications. For comparison, our qubit neural network and a conventional neural network are evaluated. As a result, we find it is possible for our qubit neural network to swing up and stabilize the inverted pendulum while not for any conventional neural network.

Noriaki Kouda, Nobuyuki Matsui, H. Nishimura +1 more

Takuya Teguri, Teijiro Isokawa, Nobuyuki Matsui +2 more

We propose a neural network model based on quantum information processing with quaternionic representation and operations, called Quaternionic Qubit Neural Network. The state of a neuron is represented by a point on the Bloch sphere with incorporating quaternionic representation. The operations for this neuron also follow the operations in quaternions. The proposed neural networks are evaluated through numerical experiments for predicting chaotic time series produced by a Lorentz system. They have better performances in predicting long-term series, as compared to conventional (real-valued) neural networks.

Noriaki Kouda, Nobuyuki Matsui, H. Nishimura +1 more

Siraj Raval

Kazuhiko Takahashi

This paper presents the servo controller design of using a quantum neural network and investigates its characteristics for control systems. A multi-layer quantum neural network that uses qubit neurons as an information processing unit is used to design three types of neural network-based servo controller: the direct controller, the parallel controller and the self-tuning controller. Computational experiments to control the nonlinear discrete-time plant are conducted in order to evaluate the learning performance and the capability of the quantum neural controller. The results of the computational experiments confirm both the feasibility and effectiveness of the quantum neural controllers.