Quantum Cognitive Map Model Based On Quantum Fuzzy Neural Network

Quantum signal processing becomes the emerging processing method in the signal processing field, because of its parallelization characteristics. The major objective of my article is finding new model for signal processing analysis based on the analysis and the basic principle of quantum information processing.For causal relationships between the concepts of FCM are either restricted certainty or limited to the qualitative probability has certain drawbacks in complex causal systems; Meanwhile, the establishment of FCM involving more than one expert, every expert has his individual knowledge and limitations. We can not fully use the competitive, redundant or complementary of the expert knowledge to make the available decision. Therefore, we propose the Quantum Cognitive Map (QCM) by extending concepts of unconventional FCM to the quantum computation.This article focuses on the construction of quantum cognitive map with training, and we have completed the following works:Firstly, we propose the Quantum Concept as the basic processing unit of QCM, which is represented by the superposition of the states. In this way, we can not only obtain the multi-state description using Q-bits providing by the expert's knowledge, but also indicate different state of the concept between weights, which retain the influence of various opinion, so it is better to simulate the real system.Secondly, the inference process of Quantum Concept has the similar property with QN, which facilitates the structure of the training model of QCM. We construct the quantum fuzzy neural network according to the definition and description of the QCM. In this way, the QCM model can represent the multi-state nature of the concept using Quantum Concept and get the exact causal relationship.Finally, the inference process generated by QCM is conceptually decomposed into interactive two jobs, namely the weight adjusted on causal relationships and dynamic update of weights. The first kind of weight illustrates the transparent interpretation for causalities in QCM, which is gained from the concept of mutual subsethood. Another dynamically allocated weight is gained by the backpropagation algorithm.In fact, as a forecasting tool, the nonlinear mapping ability of QCM is not only used to make up for the explanatory deficiencies of traditional FCM, but also enhance the system's knowledge representation and reasoning ability.