Quantum Associative Memory Based on Subspace Projection

To study the biological associative phenomena, we propose a novel quantum associative neural network model based on the subspace projection transformation of the quantum system. The patterns are encoded into a set of independent quantum state and iteratively stored into the quantum associative memory. The information retrieval is an action of projection transformation and follows a POVM measurement. Computer simulations show that it exhibits good associative ability. It can associatively recognize the face with different expression from a set of face images. In addition, the model can dramatically improve the storage capacity and process speed compared with the classical neural network model due to the quantum techniques. Compared with previous quantum associative memory models, our model can memorize and retrieval nonorthogonal quantum patterns with good performance.