Application Of Quantum Machine Learning For Encrypted Data

At present,in the face of huge computing data,parallel computing capability is particularly important.Quantum computing and cloud computing are technologies that can change the way of computing in the future.Quantum computing uses the properties of quantum physics such as coherence and entanglement to design some high-speed computing models and accelerate the classical algorithms.Because large quantum computers need certain operating conditions,it seems that it is still difficult to deploy high-performance personal quantum computers in the short term.Cloud computing can provide computing power to clients as a service.Therefore,it is necessary for the client to carry out complex quantum computing with the help of quantum cloud computing.In other words,individuals can access the quantum computer of the cloud,efficiently complete the computing tasks,return the results,and ensure its security through the quantum encryption scheme.This way will be the focus of future computing services.Quantum principal component analysis(QPCA)is a method of quantum state tomography.In this paper,QPCA is used to extract the eigenvalue matrix after the eigenvalue decomposition of the density matrix to reduce the dimension,and a quantum principal component extraction algorithm(QPCE)is proposed.Compared with the classical algorithm,this algorithm achieves an exponential acceleration ratio under certain conditions.The concrete realization of the quantum circuit is given.Compared with the classical algorithm,this algorithm achieves exponential acceleration under certain conditions,and the specific implementation of the quantum circuit is given.Considering the limited computing power of the client,the client can encrypt the quantum data and upload it to the cloud for computing,through the quantum homomorphic encryption scheme(QHE)to ensure security.A quantum homomorphic ciphertext dimensionality reduction scheme(QHEDR)is proposed.In addition,even if there is a T-gate in the calculation circuit,there is no need for interaction.When there are a large number of T-gates in the execution circuit in the quantum cloud,the T-gate update in the QHE scheme is too tedious,which causes great pressure on the client,and is not widely used in quantum machine learning algorithms.Therefore,this paper proposes a T-gate update scheme based on trusted server.In order to improve the encryption efficiency of the QHE scheme and reduce the pressure on the client.At the same time,this paper proposes a quantum k-means algorithm based on trusted server in quantum cloud computing.In this algorithm,the quantum subroutines with large amount of computation in the quantum k-means algorithm are calculated in the quantum cloud,and the ciphertext data is encrypted by the QHE scheme to ensure security.And on this basis,the T-gate update based on trusted server is applied.Compared with the original algorithm,this algorithm solves the problem that the client needs a lot of computing resources,and reduces the load of the client for the second time.The computational efficiency is improved.We have completed the experiment of quantum principal component extraction algorithm(QPCE)on IBM Qiskit and IBM Quantum Experience's real quantum computer.And the correctness of the QHE scheme is verified when there are multiple T-gate complex circuits in the quantum circuit.The experimental results show that the quantum principal component extraction algorithm proposed by us is feasible.In addition,we verify the correctness of the quantum subprograms SwapTest and GroverOptim based on trusted server in plaintext and ciphertext respectively.Experimental results show that compared with the original quantum kmeans algorithm,this algorithm has better performance in reducing client load cost and protecting cloud privacy data.