Research On The Model Of Quantum Neural Network With Quantum Architecteur And Quantum Algorithm

Quantum neural network based upon the basic theory of quantum computing and theclassical artificial neural network is a new method which can study neural network. And it hadaroused widespread attention of researchers because of the unique functions produce the hugeadvantage on some problems.Based on the analyses of the main features and shortages of quantum neural network andthe existing improved Grover’s search algorithm, three novel quantum neural networks withquantum learning are proposed in this paper.(1) Multi-patterns with high-probability based on quantum partial search algorithm isproposedBased on analysis on the Grover’s algorithm, to overcome the shortcomings of timecomplexity of existing searching the multi-patterns based upon quantum partial searchalgorithm proposed by Vladimir E.korpin, our research focuses on two algorithms, one for thesearching the multi-patterns by quantum hierarch partial search algorithm and one for directquantum partial search. The analysis of the time complexity strongly suggest that the timecomplexity of two suggested algorithm are superior to the quantum algorithm put forward byVladimir E.korpin, in addition, the second algorithm not only make that the probability ofcollapsing into the desired multi-patterns as1, but also the time complexity is less than orequal toO Nm, where N is the items of the database and m is the number of the desired multi-patterns, m≥2.(2) Neural network based on quantum architecture is proposedAt present, the existing quantum neural network models all introduce quantumcomputation based on the traditional neural network. It is difficult to solve the two keyproblems facing quantum neural networks: the nonlinearity in the neuron operation andefficient use of quantum superposition in the learning algorithm, so that researchers arealways searching for new methods. In order to solve the above two problems, this paperproposes a single neuron model based on the Boolean functions and the improved Grover’salgorithm. According to the case analysis, this model proposed in this paper pursues at a largeextent the parallelism of quantum computing and the nonlinearity problem in the neuronoperation can be indirectly solved by the Boolean functions.(3) Quantum associative neural network with nonlinear search algorithm is proposed Based on analysis on properties of quantum linear superposition, to overcome thecomplexity of existing quantum associative memory which was proposed by Ventura, a newstorage method for multiply patterns is proposed in this paper by constructing the quantumarray with the binary decision diagrams. Also, the adoption of the nonlinear search algorithmincreases the pattern recalling speed of this model which has multiply patterns toO (log2n-t2)=O (n t)time complexity, where n is the number of quantum bit and t is thequantum information of the t quantum bit. Results of case analysis show that the associativeneural network model proposed in this paper based on quantum learning is much better andoptimized than other researchers’ counterparts both in terms of avoiding the additional qubitsor extraordinary initial operators, storing patterns and improving the recalling speed.