QUANTUM NEURAL NETWORK

Abstract

A quantum neural network architecture. In one aspect, a quantum neural network trained to perform a machine learning task includes: an input quantum neural network layer comprising (i) multiple qubits prepared in an initial quantum state encoding a machine learning task data input, and (ii) a target qubit, a sequence of intermediate quantum neural network layers, each intermediate quantum neural network layer comprising multiple quantum logic gates that operate on the multiple qubits and target qubit; and an output quantum neural network layer comprising a measurement quantum gate that operates on the target qubit and provides as output data representing a solution to the machine learning task.

Description

Claims

1. (canceled)

2. A method for training a quantum neural network to perform a machine learning task, the method comprising: training the quantum neural network on multiple training examples, each training example comprising a machine learning task input paired with a known classification for the machine learning task input, the training comprising, for each training example: encoding the machine learning task input into an initial quantum state of multiple qubits of an input quantum neural network layer;processing the machine learning task input using one or more intermediate quantum neural network layers, wherein each intermediate quantum neural network layer comprises multiple quantum logic gates that operate on the multiple qubits and a target qubit that is also in the input quantum neural network layer, the processing comprising, for each intermediate quantum neural network layer and in a sequence, applying quantum logic gates for the intermediate quantum neural network layer to a current quantum state representing the multiple qubits and the target qubit;measuring the target qubit in an output quantum neural network layer to obtain an output that represents a solution to the machine learning task, wherein the output comprises a measurement result that depends on an evolved quantum state of the multiple qubits and target qubit, wherein the evolved quantum state depends on the multiple quantum logic gates that operate on the multiple qubits and target qubit in each intermediate quantum neural network layer;comparing the output to the known classification to determine one or more gate parameter adjustments; andadjusting values of the gate parameters from initial values to trained values according to the one or more gate parameter adjustments.

3. The method of claim 2, wherein each intermediate quantum neural network layer comprises (i) single qubit quantum logic gates, (ii) two qubit quantum logic gates, or (iii) both single qubit and two qubit quantum logic gates.

4. The method of claim 3, wherein the single qubit quantum gates comprise single qubit gates of the form exp (-iθXi).

5. The method of claim 3, wherein the two qubit quantum gates comprise two qubit gates of the form exp (iθZiZk).

6. The method of claim 2, wherein processing the machine learning task input using the one or more intermediate quantum neural network layers comprises mapping the encoded machine learning task input to an evolved state of the target qubit.

7. The method of claim 6, wherein mapping the encoded machine learning task input to the evolved state of the target qubit comprises applying a unitary operator to the initial quantum state, the unitary operator being parameterized by quantum logic gate parameters for the quantum logic gates.

8. The method of claim 2, wherein encoding the machine learning task input into an initial quantum state of the multiple qubits comprises setting a z-direction of each of the multiple qubits.

9. The method of claim 2, wherein measuring the target qubit to obtain an output that represents a solution to the machine learning task comprises measuring a y-direction of the target qubit.

10. The method of claim 2, wherein comparing the output to the known classification to determine one or more gate parameter adjustments comprises: calculating a loss function using the output and the known classification for the machine learning task; andperforming gradient descent to determine adjusted values of the gate parameters.

11. The method of claim 10, wherein the loss function is dependent on an evolved state of the multiple qubits and target qubit, wherein the evolved state is dependent on gate parameters for the quantum logic gates for each of the intermediate quantum neural network layers.

12. The method of claim 10, wherein the loss function is given by Losss,θ=ψθ,zsσy outψθ,zs−ys2where θ represents quantum gate parameters, ψ(θ,zs) represents an evolved quantum state of the multiple qubits and target qubit, σyout represents a measurement quantum gate, and ys represents the known classification.

13. The method of claim 2, further comprising performing regularization after processing a subset of training examples, wherein regularization comprises zero-norm or one-norm regularization.

14. The method of claim 2, wherein the machine learning task comprises a binary classification task, the machine learning task input comprises a Boolean function input {0, 1}”, and the solution to the machine learning task comprises a Boolean function output {0, 1}.

15. The method of claim 2, wherein the Boolean function input comprises a parity function input, subset parity function input, subset majority function input or logical AND function input.

16. An apparatus comprising: a quantum neural network implemented by one or more quantum processors; anda classical processor in data communication with the quantum neural network;wherein the apparatus is configured to perform operations for training the quantum neural network to perform a machine learning task, the operations comprising: training the quantum neural network on multiple training examples, each training example comprising a machine learning task input paired with a known classification for the machine learning task input, the training comprising, for each training example: encoding the machine learning task input into an initial quantum state of multiple qubits of an input quantum neural network layer;processing the machine learning task input using one or more intermediate quantum neural network layers, wherein each intermediate quantum neural network layer comprises multiple quantum logic gates that operate on the multiple qubits and a target qubit that is also in the input quantum neural network layer, the processing comprising, for each intermediate quantum neural network layer and in a sequence, applying quantum logic gates for the intermediate quantum neural network layer to a current quantum state representing the multiple qubits and the target qubit;measuring the target qubit in an output quantum neural network layer to obtain an output that represents a solution to the machine learning task, wherein the output comprises a measurement result that depends on an evolved quantum state of the multiple qubits and target qubit, wherein the evolved quantum state depends on the multiple quantum logic gates that operate on the multiple qubits and target qubit in each intermediate quantum neural network layer;comparing the output to the known classification to determine one or more gate parameter adjustments; andadjusting values of the gate parameters from initial values to trained values according to the one or more gate parameter adjustments.

17. The apparatus of claim 16, wherein each intermediate quantum neural network layer comprises (i) single qubit quantum logic gates, (ii) two qubit quantum logic gates, or (iii) both single qubit and two qubit quantum logic gates.

18. The apparatus of claim 17, wherein the single qubit quantum gates comprise single qubit gates of the form exp (-iθXi).

19. The apparatus of claim 17, wherein the two qubit quantum gates comprise two qubit gates of the form exp (iθZiZk).

20. The apparatus of claim 16, wherein processing the machine learning task input using the one or more intermediate quantum neural network layers comprises mapping the encoded machine learning task input to an evolved state of the target qubit.

21. The apparatus of claim 20, wherein mapping the encoded machine learning task input to the evolved state of the target qubit comprises applying a unitary operator to the initial quantum state, the unitary operator being parameterized by quantum logic gate parameters for the quantum logic gates.