Model And Application Of Quantum Machine Learning

Classical Machine Learning has developed rapidly in the past ten years,which has greatly promoted the development of data science,which relies on information mining.At the same time,Quantum Machine Learning has also received attention from researchers.The Quantum Machine Learning has carried on the innovation and the application to the classical machine learning algorithm,and it relies on the physical properties of quantum mechanics-parallelism and superposition-is theoretically faster than some traditional machine learning algorithms.Quantum Machine Learning focuses on use the data mining capabilities of Machine Learning to study problems in quantum mechanics or use the structure of quantum mechanics and quantum computing properties to optimize classical machine learning algorithms.The research steps of the former are to analyze the part of the problem that can be solved by classical machine learning,then convert quantum information into classical information,carry out data pre-processing,finally use machine learning to build a model,and compare and analyze the results with the results at the quantum level.The research steps of the latter are to prepare the corresponding quantum bits from the classical data information through the quantum gate circuit,use quantum bits to take advantage of the advantages of quantum computing,then construct a unitary operator system,calculate the quantum bits without changing the eigenvalues of the operators,and finally measure the quantum states of the output results to obtain the desired information.In this paper,we first introduce the color decoding of quantum topology error-correcting codes in the proposed quantum fault-tolerant computation,a Neural Network decoder is designed by using the classical Deep Learning Neural Network model in the quantum error correction scenario.The traditional method of topology error-correcting codes is slow in computation and inefficient in decoding.Machine Learning can solve this problem.In this decoder,the color code is projected onto the surface code,and the error is corrected by the surface code.The accuracy of the decoder with 300 iterations under unsupervised training is 96.5%,and the success rate of the decoder with the same flip error rate is obviously higher than that of the traditional method.Then we propose a new classical-quantum hybrid algorithm called Quantum Temporal Convolutional Network,which combines the classical Temporal Convolution Network and Quantum Convolutional Neural Network.The concept of Temporal Convolution Network widely used in Natural Language Processing and sequence prediction is extended to Quantum Machine Learning.On this basis,the attention mechanism module of classical machine learning algorithm is introduced and quantized.QTCN is implemented using a quantum programming framework and performs well at lower computational resources and Neural Network scales.QTCN model conforms to the requirements of classical-quantum hybrid algorithm,which provides a new algorithm and application scenario for Quantum Machine Learning.