Abstract

Classical deep learning algorithms have aroused great interest in both academia and industry for their utility in image recognition, language translation, decision-making problems and more. In this work, we have provided a quantum deep learning scheme based on multi-qubit entanglement states, including computation and training of neural network in full quantum process. In the course of training, efficient calculation of the distance between unknown unit vector and known unit vector has been realized by proper measurement based on the Greenberger–Horne–Zeilinger entanglement states. An exponential speedup over classical algorithms has been demonstrated. In the process of computation, quantum scheme corresponding to multi-layer feedforward neural network has been provided. We have shown the utility of our scheme using Iris dataset. The extensibility of the present scheme to different types of model has also been analyzed.

Highlights

  • Machine learning, as an interdisciplinary subject in the fields of computer science, mathematics, statistics and neuroscience, has made outstanding achievements in recent years

  • The computation of distance is exponential speedup and the reason is that the GHZ entanglement source has been used

  • When the appropriate measurement basis is chosen, we can obtain the probability connecting with the Mean square error (MSE)

Read more

Summary

Introduction

As an interdisciplinary subject in the fields of computer science, mathematics, statistics and neuroscience, has made outstanding achievements in recent years. Quantum Hopfield neural network based on the simulation of sparse Hamiltonian and the solution of linear systems of equations has been discussed to realize exponential speed up in computation and training [39]. Quantum convolutional neural network based on the reverse-direction correspondence of the multi-scale entanglement renormalization ansatz has been analyzed to realize efficient computation by exponentially reducing the number of parameters [42]. These schemes only focus on special models for particular tasks. One scheme encodes data into relative phases of wavefunctions and realizes the Baqprop principle, which corresponds to the error Back Propagation algorithm in classical neural network [43]. The compatibility of this scheme is discussed that other models of QNN can be combined with our scheme to realize full quantum process and more efficient computation

Scheme of quantum deep learning
Numerical results of training and test
Discussion and conclusion
H H V V is generated by the process of spontaneous parameter down
H H 2 V V 2 can be obtained when the photon 1 passes through the PBS
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call