DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant’s arguments, see Remarks, filed 05/15/2026 have been fully considered:
Regarding applicant's arguments filed with respect to the prior art rejections have been fully considered but they are moot. Applicant has amended the claims to recite new combinations of limitations. Please see below for new grounds of rejection, necessitated by Amendment.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1-3, 7-14, and 17-19 is/are r rejected under 35 U.S.C. 103 as being unpatentable over ZHENG et al. (US 20210399743 A1) in view of Harper et al. (“Efficient learning of quantum noise”, Dated: April 19, 2021).
Regarding claim 1.
ZHENG discloses a method comprising: training, by one or more processors, a quantum noise decoder comprising a machine- learning trained quantum error determination model using training data comprising operational data captured based at least in part on operation of a particular quantum processor (see ¶ 6, "Embodiments of this disclosure provide a fault tolerant and error correction decoding method and apparatus for a quantum circuit, and a chip, which achieves real-time fault tolerant and error correction decoding on error syndrome information of a quantum circuit when the error syndrome information is not perfect.", also Fig. 21: Neural network decoder corresponds to a machine-learning trained quantum error determination model and actual error syndrome information corresponds to operational data, also see ¶ 20"In addition, this solution equivalently transforms fault tolerant and error correction decoding into a classification problem, so that fault tolerant and error correction decoding can be performed on error syndrome information by using an efficient neural network classifier, thereby improving the fault tolerant and error correction decoding speed, and achieving real-time fault tolerant and error correction decoding.", also see ¶ 67: "The solutions provided in the embodiments of this disclosure involve application of machine learning technologies of Al in the field of quantum technologies, and specifically relate to application of the machine learning technologies in a decoding algorithm for QEC codes", also see ¶ 79: "Because a decoding process is an input/output function, a neural network may be constructed, and a correct input/output result is used to train a neural network, to learn accurately determining a position and a class of an error (supervised learning). Errors may be classified into two classes according to different neural network decoder output classes: one is a physical class, and the other is a logic class. A physical-class output model directly generates specific qubit information corresponding to occurrence of an error, that is, an error of which class occurs on a specific qubit.", also Fig 11, step 1101: Obtain actual error syndrome information of a quantum circuit, the actual error syndrome information being information obtained by performing a noisy error syndrome measurement on the quantum circuit by using a QEC code. - corresponds to capturing operational data);
generating, by the one or more processors a noise model for the particular quantum processor based on the machine-learning trained quantum error determination model (see Fig. 21, the error result information corresponds to a noise model for the particular quantum processor (based on actual error syndrome information), based on the machine-learning trained quantum error determination model ( neural network), also Fig. 16, step1103c-1 implies a noise model based on the quantum noise decoder (1102a)),
and providing, by the one or more processors, the noise model, wherein providing the noise model comprises at least one of (a) causing a graphical representation of at least a portion of the noise model to be provided via a display of a computing entity such that at least one component or parameter of the particular quantum processor is modified or changed based thereon or (b) providing at least a portion of the noise model as input associated with executable instructions for execution by a controller of the particular quantum processor or a computing entity in communication with the controller of the particular quantum processor such that at least one component or parameter of the particular quantum processor is modified or changed based thereon (see Fig. 16, step 1105 Transmit the error correction control signal to the quantum circuit" - corresponds to option b) , also see ¶ 84: "The control device 33 is configured to control the quantum circuit", also see ¶ 289: "The basic 1/0 system 2406 includes a display 2408 configured to display information and an input device 2409 configured for a user to input information, such as a mouse or a keyboard. The display 2408 and the input device 2409 are both connected to the processing unit 2401 by an input/output controller").
ZHENG do not teach generating, by the one or more processors a noise model for the particular quantum processor based on the machine-learning trained quantum error determination model, wherein the noise model is at least one of a time parameterized distribution or a space parameterized distribution of how noise is applied to calculations performed by the particular quantum processor.
Harper teaches generating, by the one or more processors a noise model for the particular quantum processor based on the machine-learning trained quantum error determination model (see page 3, “The specific methodology for estimating the global probability distribution as a GRF from the estimated marginals is discussed in the Methods and the supplementary information. Note that it is not necessary to know the topology of the GRF factor graph in advance, rather it should be possible to learn the topology of a GRF that adequately de scribes the observed probabilities from the data itself using techniques from classical machine learning [34]. We discuss this briefly in the supplementary information but it remains an open problem, although we aim to show how to learn the topology in future work.”, also see page 5, “structure learning heuristics can be applied to try to learn the most accurate and efficient GRF noise model that describes the error rates. In addition to the applications mentioned in the abstract, machine learning for fine-tuned error-correction decoders using the actual noise map of the device, quantum control for optimal gate synthesis, and noise-aware circuit compiling techniques are just some of the applications of this new method for characterizing quantum noise.”), wherein the noise model is at least one of a time parameterized distribution or a space parameterized distribution of how noise is applied to calculations performed by the particular quantum processor (see page 3, “Importantly, the observed error rates are still capable of describing many-body correlations within arbitrary sets of qubits… To achieve an efficient protocol, scaling polynomial in (n), we introduce the notion of a Gibbs random field to de scribe the error rates p(j). A Gibbs random field (GRF) is a strictly positive probability distribution that obeys certain conditional independence properties known as Markov conditions (see Methods).”, also see page 3, “FIG. 2. Results from the protocol run in single-qubit mode. (a) Spatial layout of the qubits in the 14 qubit Melbourne architecture. Edges in the schematic graph correspond to qubit pairs that can be coupled via a two-qubit gate. The factor graph below that is for a Gibbs random field (GRF) that models quantum noise via spatially local correlations in the device. Each diamond-shaped node is a factor that can describe arbitrary correlations among the qubits connected to it.”, also page 2, “Third, take a Walsh-Hadamard transform of each of the empirical probability distributions. The resulting values will exhibit an exponential decay with respect to the sequence length parameter, m, so in step 4 we fit to such a model to learn the decay constants.”, also see figure 1, description, “Each red box in the quantum circuits of step 1 denotes a sequence of random Clifford gates, and the circuits themselves are run for varying lengths m (with time moving from right to left). The final gate Cy is not random, but is rather an inversion gate returning the system to its initial state in the ideal case” also see page 6, “distance forms a bounded metric on the space of probability distributions over a given probability space and, for discrete distributions,”, also see page 14, “Use the relevant estimates of the decay parameters (with a forward Walsh Hadamard transform) to reconstruct the entire probability distribution”).
Both ZHENG and Harper pertain to the problem of quantum, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine ZHENG and Harper to teach the above limitations. The motivation for doing so would be “Noise is the central obstacle to building large-scale quantum computers. Quantum systems with sufficiently uncorrelated and weak noise could be used to solve computational problems that are intractable with current digital computers. There has been substantial progress towards engineering such systems [1–8]. However, continued progress depends on the ability to characterize quantum noise reliably and efficiently with high precision [9]. Here we describe such a protocol and report its experimental implementation on a 14-qubit superconducting quantum architecture. The method returns an estimate of the effective noise and can detect correlations within arbitrary sets of qubits. We show how to construct a quantum noise correlation matrix allowing the easy visualization of correlations between all pairs of qubits, enabling the discovery of long-range two-qubit correlations in the 14 qubit device that had not previously been detected. Our results are the first implementation of a provably rigorous and comprehensive diagnostic protocol capable of being run on state of the art devices and beyond. These results pave the way for noise metrology in next-generation quantum devices, calibration in the presence of crosstalk, bespoke quantum error-correcting codes [10], and customized fault-tolerance proto cols [11] that can greatly reduce the overhead in a quantum computation.” (see Harper abstract).
Regarding claim 2.
ZHENG and Harper teaches the method of claim 1,
ZHENG further discloses wherein the operational data comprises calibration data generated through operation of the particular quantum processor (see Fig. 4 discloses periodic calibration and error correction).
Regarding claim 3.
ZHENG and Harper teaches the method of claim 2,
ZHENG further discloses wherein the calibration data is captured periodically during operation of the particular quantum processor (see Fig. 4 discloses periodic calibration and error correction).
Regarding claim 7.
ZHENG and Harper teaches the method of claim 1,
ZHENG further discloses wherein the quantum noise decoder comprises a noise model generation module configured to generate the noise model for the particular quantum processor based at least in part on output of the machine-learning trained quantum error determination model (Fig. 18, Sub-step 18th: Perform OPT on the sample quantum circuit, to
extract a noise model of the sample quantum circuit, the noise model being configured to generate the data error and the measurement error through simulation. - Fig. 21 shows that the noise model is generated based on the output of the machine-learning error determination model (neural network)).
Regarding claim 8.
ZHENG and Harper teaches the method of claim 1,
ZHENG further discloses wherein the at least one component or parameter is part of or used by a real-time quantum error decoder to correct for quantum errors during operation of the particular quantum processor (see ¶ 285: "In addition, this solution equivalently transforms fault tolerant and error correction decoding into a classification problem, so that fault tolerant and error correction decoding can be performed on error syndrome information by using an efficient neural network classifier, thereby improving the fault tolerant and error correction decoding speed, and achieving real-time fault tolerant and error correction decoding. If a suitable neural network classifier is selected, the speed of the decoding algorithm can be greatly improved, thereby paving the way for implementation of real-time fault tolerant and error correction decoding." - Fig 4, Error correction control signal- Fig. 23: error correction signal generation module 2240).
Regarding claim 9.
ZHENG and Harper teaches the method of claim 1,
ZHENG further discloses wherein at least one component or parameter is a hardware component or a physical parameter of the particular quantum processor (see ¶ 0079: "Because a decoding process is an input/output function, a neural network may be constructed, and a correct input/output result is used to train a neural network, to learn accurately determining a position and a class of an error (supervised learning). Errors may be classified into two classes according to different neural network decoder output classes: one is a physical class, and the other is a logic class. A physical-class output model directly generates specific qubit information corresponding to occurrence of an error, that is, an error of which class occurs on a specific qubit.", Fig 3 and its description in §0082 - §0084 : "The quantum circuit 31 is a circuit acting on physical qubits. ( .. .) The control device 33 is configured to control the quantum circuit ( ... .) control device 33 transforms the instruction into an electronic/microwave control signal and inputs the electronic/ microwave control signal into the dilution refrigerator 32, to control superconducting qubits at 10 mK").
Regarding claim 10.
ZHENG and Harper teaches the method of claim 1,
ZHENG further discloses wherein the at least one component or parameter corresponds to a re-calibration of a hardware component of the particular quantum processor or a software process of the controller of the particular quantum processor (see ¶ 0079: "Because a decoding process is an input/output function, a neural network may be constructed, and a correct input/output result is used to train a neural network, to learn accurately determining a position and a class of an error (supervised learning). Errors may be classified into two classes according to different neural network decoder output classes: one is a physical class, and the other is a logic class. A physical-class output model directly generates specific qubit information corresponding to occurrence of an error, that is, an error of which class occurs on a specific qubit.", Fig 3 and its description in §0082 - §0084 : "The quantum circuit 31 is a circuit acting on physical qubits. ( .. .) The control device 33 is configured to control the quantum circuit ( ... .) control device 33 transforms the instruction into an electronic/microwave control signal and inputs the electronic/ microwave control signal into the dilution refrigerator 32, to control
superconducting qubits at 10 mK").
Regarding claim 11.
ZHENG and Harper teaches the method of claim 1,
ZHENG further discloses wherein the quantum noise decoder is a real-time quantum error decoder configured to cause correction of quantum errors during operation of the particular quantum processor (see ¶ 285: "this solution equivalently transforms fault tolerant and error correction decoding into a classification problem, so that fault tolerant and error correction decoding can be performed on error syndrome information by using an efficient neural network classifier, thereby improving the fault tolerant and error correction decoding speed, and achieving real-time fault tolerant and error correction decoding. If a suitable neural network classifier is selected, the speed of the decoding algorithm can be greatly improved, thereby paving the way for implementation of real-time fault tolerant and error correction decoding." - Fig 4, Error correction control signal - Fig. 23: error correction signal generation module 2240).
Regarding claim 12.
ZHENG and Harper teaches the method of claim 1,
ZHENG further discloses wherein the noise model characterizes noise present in the operational data for the particular quantum processor (see ¶ 178: "In another possible implementation, quantum process tomography (QPT) may be first experimentally performed on qubits, to extract an actual noise model.").
Claims 13-14, 17-19 recites an apparatus to perform the method recited in claims 1-3, 10-12. Therefore the rejection of claims 1-3, 10-12 above applies equally here. ZHENG also teaches the addition elements of claim 13 not recited in claim 1 comprising at least one non-transitory memory storing computer-executable instructions and a processing device (see ¶ 288, “a processing unit 2401 (for example, a central processing unit (CPU) and/or a graphics processing unit (GPU)), a system memory 2404 including a random access memory (RAM) 2402 and a read-only memory (ROM) 2403, and a system bus 2405 connecting the system memory 2404 to the processing unit 2401.”).
Claim(s) 4 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over ZHENG et al. (US 20210399743 A1) in view of Harper et al. (“Efficient learning of quantum noise”, Dated: April 19, 2021) in view of Daraeizadeh et al. (US 20220199888 A1).
Regarding claim 4.
ZHENG and Harper teaches discloses the method of claim 1,
ZHENG and Harper do not teach the limitation of claim 4.
Daraeizadeh further discloses wherein the operational data comprises spectator objects data captured through direct or indirect observation of one or more spectator objects controlled by the particular quantum processor, the one or more spectator objects controlled independently of a quantum algorithm being executed by the particular quantum processor (0074-0095: ''Apparatus and Method for Automatic Real-Time Calibration of Qubit Chips", §0078: "One embodiment of the invention implements machine-learning techniques to learn and calibrate qubit control parameters in real-time, during system operation (e.g., during execution of the underlying quantum algorithm)" - §0083: "For example, measurement discrimination units (MDUs) may determine the state of ancilla qubits, which can be used to determine the state of corresponding data qubits in the quantum processor 207." – where the state of ancilla qubits correspond to spectator data).
ZHENG, Harper and Daraeizadeh pertain to the problem of quantum, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine ZHENG, Harper and Daraeizadeh to teach the above limitations. The motivation for doing so would be “a qubit measurement unit to measure one or more sensors associated with a corresponding one or more of the qubits of the plurality of qubits to produce one or more corresponding measured values; and a machine-learning engine to evaluate the one or more measured values in accordance with a machine-learning process to generate updated control parameters, wherein the quantum controller is to use the updated control parameters to generate subsequent sequences of EM pulses to manipulate the states of the plurality of qubits.” (see Daraeizadeh abstract).
Claim 20 recites an apparatus to perform the method recited in claim 4. Therefore the rejection of claim 4 above applies equally here.
Claim(s) 5-6 and 15-16 is/are rejected under 35 U.S.C. 103 as being unpatentable over ZHENG et al. (US 20210399743 A1) in view of Harper et al. (“Efficient learning of quantum noise”, Dated: April 19, 2021) in view of BRACCIA et al. ("Quantum Noise Sensing by generating Fake Noise", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 19 July 2021 (2021-07-19), XP091013442).
Regarding claim 5.
ZHENG and Harper teaches the method of claim 1,
ZHENG and Harper do not teach the limitation of claim 5.
BRACCIA discloses wherein the quantum noise decoder comprises a generative adversarial network (GAN) comprising a generator and a discriminator and the generator is configured to generate simulated operational data (see Abstract: "we propose a very promising framework to characterize noise in a realistic quantum device, even in the case of spatially and temporally correlated noise (memory channels) affecting quantum circuits. The key idea is to learn about the noise by mimicking it in a way that one cannot distinguish between the real (to be sensed) and the fake (generated) one. (. .. ) We believe our SuperQGANs pave the way for new hybrid quantum classical machine learning protocols for a better characterization and control of the current and future unavoidably noisy quantum devices." - Fig 1 and Fig 2, page 8: "In this work we generalize standard Quantum Generative Adversarial Networks for quantum state learning to what we here call SuperQGANs, i.e. a quantum machine learning tool to learn and reproduce quantum maps.").
ZHENG, Harper and BRACCIA pertain to the problem of quantum, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine ZHENG, Harper and BRACCIA to teach the above limitations. The motivation for doing so would be “we propose a very promising framework to characterize noise in a realistic quantum device, even in the case of spatially and temporally correlated noise (memory channels) affecting quantum circuits. The key idea is to learn about the noise by mimicking it in a way that one cannot distinguish between the real (to be sensed) and the fake (generated) one. (. .. ) We believe our SuperQGANs pave the way for new hybrid quantum classical machine learning protocols for a better characterization and control of the current and future unavoidably noisy quantum devices.” (see BRACCIA abstract).
Regarding claim 6.
ZHENG, Harper and BRACCIA teach the method of claim 5,
BRACCIA further teaches wherein the discriminator comprises or is in communication with the machine-learning trained quantum error determination model (see Abstract: "we propose a very promising framework to characterize noise in a realistic quantum device, even in the case of spatially and temporally correlated noise (memory channels) affecting quantum circuits. The key idea is to learn about the noise by mimicking it in a way that one cannot distinguish between the real (to be sensed) and the fake (generated) one. (. .. ) We believe our SuperQGANs pave the way for new hybrid quantum classical machine learning protocols for a better characterization and control of the current and future unavoidably noisy quantum devices." - Fig 1 and Fig 2, page 8: "In this work we generalize standard Quantum Generative Adversarial Networks for quantum state learning to what we here call SuperQGANs, i.e. a quantum machine learning tool to learn and reproduce quantum maps.").
The motivation utilized in the combination of claim 5, super, applies equally as well to claim 6.
Claims 15-16 recites an apparatus to perform the method recited in claims 5-6. Therefore the rejection of claims 5-6 above applies equally here.
Related prior arts:
Pyzer-Knapp et al. (US 20230012699 A1) teaches providing a defined parameter, determining whether to employ the defined parameter for a variational quantum algorithm, and running the variational quantum algorithm on a quantum system, are provided.
GAMBLE et al. (US 20230090148 A1) teaches providing a trial control-parameter value (52) to a quantum computer. Result (54) of a characterization experiment (44) enacted is received from the quantum computer according to the trial control-parameter value. Decoder estimate (56) of objective function evaluated is computed at the trial control-parameter value.
Zlokapa et al. (“A deep learning model for noise prediction on near-term quantum devices”, 2020) teaches Quantifying the output noise of a circuit C relative to a device can be done in a number of ways. One approach is to compute the total variation distance between the ideal output distribution of C (for some fixed input, say |00...0) and the output distribution obtained from running the circuit. For general circuits this would scale exponentially in the number of qubits, so we instead consider only circuits of the form UU†, for some unitary U.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IMAD M KASSIM whose telephone number is (571)272-2958. The examiner can normally be reached 10:30AM-5:30PM, M-F (E.S.T.).
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michael J. Huntley can be reached at (303) 297 - 4307. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/IMAD KASSIM/ Primary Examiner, Art Unit 2129
Prosecution Insights