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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 07/03/2025 was filed before the mailing date of the first office action. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Status of Claims
This action is in response to the amendments filed 01/05/2026. Claims 1, 8, and 14 have been amended, claims 5 and 7 have been cancelled. Claims 1-4, 6, and 8-25 are currently pending.
Response to Arguments
Claims 5 and 7 have been cancelled, therefore the rejections of claims 5 and 7 no longer stand.
Applicant’s arguments regarding the prior art rejection have been fully considered but are moot because of the new ground(s) of rejection. Applicant argues that the prior art does not teach “including a dimension of the quantum computation model used in addition to or in place of a quantum Fisher information spectrum”. Examiner notes that the Abbas reference has been brought in to teach utilizing dimensional information of a quantum model in addition to quantum Fisher information spectrum information. The prior art rejections have been updated to include the amended limitations and to clarify the reasoning given for the limitations that were not amended.
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.
Claims 1-4, 8-18, and 20-25 are rejected under 35 U.S.C. 103 as being unpatentable over Cerezo et al* (“Variational quantum algorithms”, herein Cerezo) in view of Dalli et al (US 20220114417 A1, herein Dalli), in further view of Abbas et al* (“The power of quantum neural networks”, herein Abbas).
*a copy of this document was provided in the IDS dated 05/26/2022; therefore a copy has not been provided with this action
Regarding claim 1, Cerezo teaches a method of generating a classical model to simulate a quantum computational model, the method comprising: inputting into a quantum computational model a dataset, the quantum computational model being implemented on a quantum computer; computing output results with the quantum computational model using the quantum computer; computing updated output results of the quantum computational model [based on the variation of the at least the portion of the dataset] using the quantum computer (the description of figure 1 recites “The inputs to a VQA are: a cost function C(θ), with θ a set of parameters that encodes the solution to the problem, an ansatz whose parameters are trained to minimize the cost, and (possibly) a set of training data used during the optimization. At each iteration of the loop one uses a quantum computer to efficiently estimate the cost (or its gradients)” (i.e., inputting a dataset into an iterative quantum computation model and outputting results from the quantum computation model));
and generating a classical twin model of the quantum computational model based on a relationship of the output results and the updated output results to the dataset from the quantum computational model (the description of figure 1 recites “At each iteration of the loop one uses a quantum computer to efficiently estimate the cost (or its gradients). This information is fed into a classical computer that leverages the power of optimizers to navigate the cost landscape C(θ) and solve the optimization problem in (EQ1). Once a termination condition is met, the VQA outputs an estimate of the solution to the problem” (i.e., generating a classical model corresponding to the quantum computation model based on the output of the quantum computation model)).
and wherein the generating the classical twin model comprises introducing interaction terms between two or more variables in the classical twin model to simulate entanglement in the quantum computational model (section III G 3 recites “Characterizing entanglement is crucial for understanding condensed matter systems, and the entanglement spectrum has proven to be useful in studying topological order. Since the entanglement spectrum can be viewed as the principal components of a reduced density matrix, algorithms for PCA can be used for this purpose, including the VQAs discussed before. These algorithms could potentially characterize the entanglement (and for example, topological order) in a ground state that was prepared by VQE, and hence different VQAs can be used together in a complementary manner” (i.e., the quantum-classical model can simulate entanglement));
wherein quantum information measures are used to inform the development of the classical twin model (section III G 4 recites “one prepares a probe state with variational parameters, probes the magnetic field with physical noises, measures quantum Fisher information (QFI) as a cost function, and updates the parameters to maximize it. Note that since QFI cannot be efficiently computed, an approximation of QFI can be heuristically found by optimizing the measurement basis, or by computing upper and lower bounds on the QFI” (i.e., quantum information measures can be used to develop quantum-classical twin models)).
However, while Cerezo teaches an iterative variational quantum algorithm (see at least figure 1 and its description), Cerezo does not explicitly teach introducing a variation to at least a portion of the dataset into the quantum computer.
Dalli teaches introducing a variation to at least a portion of the dataset into the quantum computer (para. [0088] recites “The interaction and moderator component 1513 is a component of the EIGS (i.e., explanation and interpretation generation system) neurosymbolic architecture, and may contain information about the statistical correlations and causal interactions identified in the model 904, the EIGS and its components”. Para. [0090] recites “It may be contemplated that interactions and moderators component 1513 implements a suitable resampling method, applicable to any relevant piece of information identified in the model 904”. Para. [0091] recites “The EIGS may implement moderators as a categorical or a quantitative variable that affects the direction and/or strength of the relation between the Interactions identified in interactions and moderators component 1513”. Para. [0166] recites “an EIGS may be implemented using a quantum processing system. It is contemplated that such a Quantum EIGS will have characteristics that are similar to an EIGS implemented on a classical non-quantum processing system with the addition of quantum specific extensions” (i.e., resampling, or introducing variations to a dataset)).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine these teachings by supplementing the variational quantum algorithm from Cerezo with the additional data variation methods from Dalli. Cerezo and Dalli are both directed to quantum circuit simulation and implementation programs. One of ordinary skill in the art would be motivated to add the variation and explanatory methods from Dalli to further enhance the interpretability of the variational quantum algorithm from Cerezo.
However, while Cerezo teaches that an effective dimension can be sued to measure the capacity of a quantum neural network in at least section V C 2, and a quantum Fisher information measurement method in at least section III G 4, the combination of Cerezo and Dalli does not explicitly teach including a dimension of the quantum computation model used in addition to or in place of a quantum Fisher information spectrum.
Abbas teaches including a dimension of the quantum computation model used in addition to or in place of a quantum Fisher information spectrum (section I para. 4 recites “We therefore, turn our attention to measures that are calculable in practice and incorporate the distribution of data. In particular, measures such as the effective dimension have been motivated from an information-theoretic standpoint and depend on the Fisher information; a quantity that describes the geometry of a model’s parameter space and is essential in both statistics and machine learning. We argue that the effective dimension is a robust capacity measure through proof of a novel generalization bound with supporting numerical analyses, and use this measure as a tool to study the power of quantum and classical neural networks”. Section I para. 6-7 recite “we examine the trainability of quantum and classical neural networks by analyzing the Fisher information matrix, which is incorporated by the effective dimension. In this way, we can explicitly relate the effective dimension to model trainability. We find that well-designed quantum neural networks are able to achieve a higher capacity and faster training ability than comparable classical feedforward neural networks. Capacity is captured by the effective dimension, whilst trainability is assessed by leveraging the information-theoretic properties of the Fisher information” (i.e., including dimension information from a quantum model in addition to Fisher information spectrum measurement data)).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine these teachings by modifying the quantum Fisher information measurement method from Cerezo with the effective dimension information from Abbas. Cerezo and Abbas are both directed to methods of measuring information associated with a quantum machine learning model, and section V C 2 of Cerezo states “it has been shown that quantum neural networks can achieve a significantly higher capacity, as measured by the effective dimension, than comparable classical neural networks”. One of ordinary skill in the art would be motivated to modify the quantum Fisher information measurement from Cerezo with the effective dimension information from Abbas to more accurately reflect the capacity of the quantum machine learning model.
Regarding claim 2, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, further comprising determining variable importance scores from the classical twin model based on a likelihood of a change in data outcome depending on a change of an input data point (Dalli para. [0157] recites “The EIGS (i.e., explanation and interpretation generation system) classifies user-centric explanation types as follows: (i.) how-type explanations that represent how the model 904 works, via model interpretability and optional justification 9063; (ii.) why-type explanations that explain why a particular explanation and/or prediction has been output for the particular input query 902, with either a model agnostic or model dependent explanation that may involve cause-and-effect explanation chains; (iii.) why-not (contrastive) explanations that explain why a specific output was not in the output of the EIGS system and explain the reasons for differences between a model prediction and the system user expected outcome; (iv.) what-if explanations that explain how different inputs affect the model output and may be either automatically recommended by the EIGS or be chosen interactively for exploration through plan and question component 1537 or interactive context component 1544”. Para. [0259] recites “The same input and output pairs may be presented to the relevant XAI model being evaluated, which may produce an explanation Es for each input-output pair. Some suitable distance or scoring function Sh may compare a human explanation Eh with Es, giving an evaluation function Vh of the form: Vh(
I
,
O
,
E
s
,
I
,
O
,
E
h
) = Sh (Es, Eh)” (i.e., determining how important different inputs, or variations to variables in a dataset, are based on the likelihood that those inputs affect the model output)).
Regarding claim 3, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 2, further comprising updating the classical twin model based on the variable importance scores (Dalli teaches determining important variables in at least paragraphs [0157] and [0259]. Cerezo teaches updating the quantum-classical model using an iterative optimization method in at least figure 1 and its description)).
Regarding claim 4, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, wherein computing output results with the quantum computational model using the quantum computer comprises encoding data, processing data, measuring for quantum kernel calculation, and estimating of prediction and cost function (Cerezo section II para. 2 recites “As schematically shown in Fig. 1, the first step to developing a VQA is to define a cost (or loss) function C which encodes the solution to the problem. One then proposes an ansatz, that is, a quantum operation depending on a set of continuous or discrete parameters θ that can be optimized (see below for a more in-depth discussion of ansatzes). This ansatz is then trained in a hybrid quantum-classical loop to solve the optimization task” (i.e., encoding and processing data using a quantum computer). Cerezo section II A para. 2 recites “one must be able to ‘efficiently estimate’ C(θ) by performing measurements on a quantum computer and possibly performing classical postprocessing” (i.e., measuring a quantum calculation and estimating the prediction and cost function using a quantum computer)).
Regarding claim 8, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, further comprising assessing the twin classical model using a weighted combination of metrics (Cerezo section III A 4 recites “The main idea behind subspace Variational Quantum Eigensolver is to train a unitary for preparing states in the lowest energy subspace of H. There are two variants of subspace VQE called weighted and non-weighted subspace VQE. For the weighted subspace VQE, one considers a cost function with ordered weights and easily prepared mutually-orthogonal states. By minimizing the cost function, one approximates the subspace of the lowest eigenstates” (i.e., applying weighted metrics to the variational quantum eigensolver, which is an application of a variational quantum algorithm)).
Regarding claim 9, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, further comprising determining a chaotic behavior or sensitivity of the updated output results of the quantum computational model based on the variation of the at least the portion of the dataset (Cerezo section II D 1 para. 2 recites “A different gradient-based approach is based on simulating an imaginary time evolution, or equivalently by using the quantum natural gradient descent method, which is based on notions of information geometry. Natural gradient descent works instead on a space with a metric tensor that encodes the sensitivity of the quantum state to variations in the parameters. Using this metric tensor, typically accelerates the convergence of the gradient update steps, allowing a given level of precision to be attained with fewer iterations” (i.e., determining a sensitivity of the output of the variational quantum computation model)).
Regarding claim 10, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, wherein introducing the variation to the at least the portion of the dataset into the quantum computer comprises using a contrastive explainability algorithm (Dalli para. [0157] recites “The EIGS (i.e., explanation and interpretation generation system) classifies user-centric explanation types as follows: (i.) how-type explanations that represent how the model 904 works, via model interpretability and optional justification 9063; (ii.) why-type explanations that explain why a particular explanation and/or prediction has been output for the particular input query 902, with either a model agnostic or model dependent explanation that may involve cause-and-effect explanation chains; (iii.) why-not (contrastive) explanations that explain why a specific output was not in the output of the EIGS system and explain the reasons for differences between a model prediction and the system user expected outcome; (iv.) what-if explanations that explain how different inputs affect the model output and may be either automatically recommended by the EIGS or be chosen interactively for exploration through plan and question component 1537 or interactive context component 1544” (i.e., a contrastive explainability algorithm used to show how different inputs, or variations to a dataset, affect the model output)).
Regarding claim 11, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, wherein computing updated output results of the quantum computational model using the quantum computer comprises calculating a variation of an updated output result relative to a variation of a data point selected from the at least portion of the dataset (the description of figure 1 of Cerezo recites “the ansatz is shown as a parameterized quantum circuit (on the left), which is analogous to a neural network (also shown schematically on the right). At each iteration of the loop one uses a quantum computer to efficiently estimate the cost (or its gradients). This information is fed into a classical computer that leverages the power of optimizers to navigate the cost landscape C(θ) and solve the optimization problem in (EQ1)” (i.e., figure 1 depicts an iterative variational quantum-classical model)).
Regarding claim 12, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, wherein the quantum computational model comprises a computational pipeline having two or more computational steps, and at least one quantum computational step and at least one classical step (the description of figure 1 of Cerezo recites “the ansatz is shown as a parameterized quantum circuit (on the left), which is analogous to a neural network (also shown schematically on the right). At each iteration of the loop one uses a quantum computer to efficiently estimate the cost (or its gradients). This information is fed into a classical computer that leverages the power of optimizers to navigate the cost landscape C(θ) and solve the optimization problem in (EQ1)” (i.e., figure 1 depicts a computation pipeline having at least one quantum computation step and at least one classical step)).
Regarding claim 13, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, wherein inputting into the quantum computational model the dataset comprises inputting into the quantum computational model a classical dataset or a quantum dataset (the description of figure 1 of Cerezo recites “The inputs to a VQA are: a cost function C(θ), with θ a set of parameters that encodes the solution to the problem, an ansatz whose parameters are trained to minimize the cost, and (possibly) a set of training data used during the optimization” (i.e., figure 1 depicts inputting a dataset into the quantum computation model)).
Claim 14 is a system claim and its limitation is included in claim 1. The only difference is that claim 14 requires a system. Therefore, claim 14 is rejected for the same reasons as claim 1.
Claim 15 is a system claim and its limitation is included in claim 2. Claim 15 is rejected for the same reasons as claim 2.
Claim 16 is a system claim and its limitation is included in claim 3. Claim 16 is rejected for the same reasons as claim 3.
Claim 17 is a system claim and its limitation is included in claim 4. Claim 17 is rejected for the same reasons as claim 4.
Claim 18 is a system claim and its limitation is included in claim 5. Claim 18 is rejected for the same reasons as claim 5.
Claim 20 is a system claim and its limitation is included in claim 7. Claim 20 is rejected for the same reasons as claim 7.
Claim 21 is a system claim and its limitation is included in claim 8. Claim 21 is rejected for the same reasons as claim 8.
Claim 22 is a system claim and its limitation is included in claim 9. Claim 22 is rejected for the same reasons as claim 9.
Claim 23 is a system claim and its limitation is included in claim 10. Claim 23 is rejected for the same reasons as claim 10.
Claim 24 is a system claim and its limitation is included in claim 11. Claim 24 is rejected for the same reasons as claim 11.
Claim 25 is a system claim and its limitation is included in claim 13. Claim 25 is rejected for the same reasons as claim 13.
Claims 6 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Cerezo et al* (“Variational quantum algorithms”, herein Cerezo) in view of Dalli et al (US 20220114417 A1, herein Dalli), in further view of Abbas et al* (“The power of quantum neural networks”, herein Abbas), in further view of Greco et al (US 20180225357 A1, herein Greco).
Regarding claim 6, the combination of Cerezo, Dalli, and Abbas teaches the method according to claim 1, and introducing a variation to the at least the portion of the dataset into the quantum computer.
However, the combination of Cerezo, Dalli, and Abbas does not teach introducing a variation to a broader portion of the dataset and then iteratively narrowing the broader portion of the dataset.
Greco teaches introducing a variation to a broader portion of the dataset and then iteratively narrowing the broader portion of the dataset (para. [0034] recites “classification program 114 may define a broad “Faulty Gear” category in early iterations, and classification program 114 may define more narrow “Stripped Gear,” “Cracked Gear,” and “Broken Gear” categories in later iterations” (i.e., broad portions of a dataset can be used in earlier iterations and narrower portions of a dataset can be used in later iterations of a model).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine these teachings by utilizing the broad to narrow classification approach taught by Greco to modify the variations taught by the VQA taught by Cerezo (as modified by Dalli and Abbas). Cerezo teaches that its VQA may be applied as a classifier in at least section III F 1; as such, one of ordinary skill would recognize that the methods of iteratively classifying data in a broad narrow manner from Greco could be applied to the input data used by the variational quantum classifier from Cerezo.
Claim 19 is a system claim and its limitation is included in claim 6. Claim 19 is rejected for the same reasons as claim 6.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
“Variational Quantum Algorithm for Estimating the Quantum Fisher Information” (Beckey et al) teaches a method for estimating the lower and upper bounds of a quantum Fisher information estimation and variationally prepare a state that maximizes the quantum Fisher information.
US 20220398460 A1 (Dalli et al) teaches an explanation and interpretation generation system (EIGS) for optimizing explainable quantum models based on features, or dimensions of the quantum models.
US 20220292377 A1 (Benedetti et al) teaches a method for utilizing variational inference methods in a hybrid quantum-classical model to monitor and control a physical process.
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 LEAH M FEITL whose telephone number is (571) 272-8350. The examiner can normally be reached on M-F 0900-1700 EST.
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, Viker Lamardo can be reached on (571) 270-5871. 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.
/L.M.F./ Examiner, Art Unit 2147
/VIKER A LAMARDO/Supervisory Patent Examiner, Art Unit 2147