# US 12,001,918 B2 · Classification using quantum neural networks > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,001,918 B2 - Document ID: US12001918B2 - Status: Granted - Kind: B2 - Application number: 16/962,348 - Filed: 2019-01-16 - Published: 2020-10-29 - Granted: 2024-06-04 - Earliest claimed: 2018-01-18 - Assignees: Google Inc. - Applicants: Google LLC - Inventors: Edward Henry Farhi; Hartmut Neven - Web page: https://patentgrep.com/patent/US12001918B2 ## CPC - G06N 10/60: Quantum algorithms, e.g. based on quantum optimisation, quantum Fourier or Hadamard transforms - G06N 3/063: using electronic means - G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections - G06N 3/084: Backpropagation, e.g. using gradient descent - G06N 3/09: Supervised learning - G06N 10/40: Physical realisations or architectures of quantum processors or components for manipulating qubits, e.g. qubit coupling or qubit control - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract This disclosure relates to classification methods that can be implemented on quantum computing systems. According to a first aspect, this specification describes a method for training a classifier implemented on a quantum computer, the method comprising: preparing a plurality of qubits in an input state with a known classification, said plurality of qubits comprising one or more readout qubits; applying one or more parameterised quantum gates to the plurality of qubits to transform the input state to an output state; determining, using a readout state of the one or more readout qubits in the output state, a predicted classification of the input state; comparing the predicted classification with the known classification; and updating one or more parameters of the parameterised quantum gates in dependence on the comparison of the predicted classification with the known classification. ## Claims ### Claim 1 A method for training a classifier implemented on a quantum computer, the method comprising: preparing a plurality of qubits in an input state with a known classification, said plurality of qubits comprising one or more data qubits that represent a state to be classified and one or more readout qubits; applying one or more layers of parameterised quantum gates to the plurality of qubits to transform the input state to an output state, wherein each layer of parameterised quantum gates comprises a plurality of two-qubit quantum gates, each of the two-qubit quantum gates acting between a readout qubit and a different one of the data qubits; determining, using a readout state of the one or more readout qubits in the output state, a predicted classification of the input state; comparing the predicted classification with the known classification; and updating, during a training process, one or more trained parameters of the parameterised quantum gates in dependence on the comparison of the predicted classification with the known classification. ### Claim 2 (depends on 1) The method of claim 1, further comprising iterating until one or more threshold conditions are met. ### Claim 3 (depends on 2) The method of claim 2, further comprising determining the readout state of the one or more readout qubits, wherein determining the readout state comprises repeatedly: preparing the plurality of qubits in the input state; applying the parameterised quantum gates to the input state; and measuring the readout state of the one or more readout qubits. ### Claim 4 (depends on 1) The method of claim 1, wherein comparing the predicted classification with the known classification comprises determining an estimated sample loss. ### Claim 5 (depends on 4) The method of claim 4, wherein updating the one or more trained parameters comprises modifying a parameter to decrease the estimated sample loss. ### Claim 6 (depends on 5) The method of claim 5, comprising modifying the one or more trained parameters using a gradient descent method. ### Claim 7 (depends on 1) The method of claim 1, wherein the parameterised quantum gates each implement a one parameter unitary transformation. ### Claim 8 (depends on 7) The method of claim 7, wherein each of the parameterised quantum gates comprises one of: a single qubit quantum gate; a two qubit quantum gate; or a three qubit quantum gate. ### Claim 9 (depends on 7) The method of claim 7, wherein one or more of the quantum gates implements a unitary transformation of the form: exp(iθΣ), where θ is a parameter parameterising the quantum gate and Σ is a generalised Pauli operator acting on one or more of the plurality of qubits. ### Claim 10 (depends on 1) The method of claim 1, wherein the input state comprises a superposition of binary strings. ### Claim 11 (depends on 1) The method of claim 1, wherein the input state comprises an arbitrary quantum state. ### Claim 12 (depends on 1) The method of claim 1, further comprising using a classical artificial neural network to prepare the input state, comprising: inputting an initial state to be classified into the classical artificial neural network; and providing an output of the classical artificial neural network to the quantum computer as the input state with the known classification. ### Claim 13 (depends on 1) The method of claim 1, comprising determining the predicted classification of the input state from the readout state using a classical artificial neural network, comprising: inputting the readout state into the classical artificial neural network, wherein the classical artificial neural network is trained to determine, using the readout state, the predicted classification of the input state. ### Claim 14 (depends on 1) The method of claim 1, further comprising: applying the one or more parameterised quantum gates to transform a plurality of qubits from an unclassified input state to a classifying output state; determining a readout state from measurements on one or more readout qubits in the classifying output state; and classifying the unclassified input state in dependence on the readout state. ### Claim 15 A method of classification performed using a quantum computer, the method comprising: applying one or more layers of parameterised unitary operators to transform a plurality of qubits from an input state to an output state, wherein the plurality of qubits comprise one or more data qubits that represent the input state to be classified and one or more readout qubits; and each layer of parameterised unitary operators comprises a plurality of two-qubit quantum gates, each of the two-qubit quantum gates acting between a readout qubit and a different one of the data qubits, wherein parameters of the one or more parameterised unitary operators comprise trained parameters; determining a readout state from measurements on the one or more readout qubits in the output state; and classifying the input state in dependence on the readout state. ### Claim 16 A quantum computing system comprising: a plurality of qubits; and one or more parameterised quantum gates, wherein the system is configured to: prepare the plurality of qubits in an input state with a known classification, said plurality of qubits comprising one or more data qubits that represent a state to be classified and one or more readout qubits; apply one or more layers of the one or more parameterised quantum gates to the plurality of qubits to transform the input state to an output state, wherein each layer of parameterised quantum gates comprises a plurality of two-qubit quantum gates, each of the two-qubit quantum gates acting between a readout qubit and a different one of the data qubits; determine, using a readout state of the one or more readout qubits in the output state, a predicted classification of the input state; compare the predicted classification with the known classification; and update, during a training process, one or more trained parameters of the parameterised quantum gates in dependence on the comparison of the predicted classification with the known classification. ### Claim 17 A quantum computing system comprising: a plurality of qubits; and one or more parameterised quantum gates, wherein the system is configured to: apply the one or more parameterised quantum gates to transform the plurality of qubits from an unclassified input state to an classifying output state, trained parameters of the one or more parameterised quantum gates determined using a classifier trained using a method comprising: preparing a plurality of qubits in an input state with a known classification, said plurality of qubits comprising one or more data qubits that represent a state to be classified and one or more readout qubits; applying one or more layers of the one or more parameterised quantum gates to the plurality of qubits to transform the input state to an output state, wherein each layer of parameterised quantum gates comprises a plurality of two-qubit quantum gates, each of the two-qubit quantum gates acting between a readout qubit and a different one of the data qubits; determining, using a readout state of the one or more readout qubits in the output state, a predicted classification of the input state; comparing the predicted classification with the known classification; and updating, during a training process, one or more trained parameters of the parameterised quantum gates in dependence on the comparison of the predicted classification with the known classification; determine a readout state from measurements on one or more readout qubits in the classifying output state; and classify the unclassified input state in dependence on the readout state.