DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application is being examined under the pre-AIA first to invent provisions.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 07/27/2022 and 08/14/2025 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Arguments
103 Rejection Arguments
Applicant asserts:
Applicant argues, on page 15-16, that “Claim 1 requires that "at least one processor... analyzes metadata of a trained quantum machine learning model submitted by a user for execution for performing a defined task, wherein the metadata comprises a first coupling map generated during training of the trained quantum machine learning model on a first quantum computing system, and wherein the first coupling map comprises a first connection topology of physical qubits that were employed of the first quantum computing system for the training of the trained quantum machine learning model." None of Jay, Dou, or Glushakov discloses metadata of a trained quantum machine learning (QML) model that includes such a training-time coupling map… nor as a record of which actual physical qubits were actually used during training on a first device.”
Examiner response:
Applicant’s arguments with respect to claim(s) 1 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant asserts:
Applicant argues, on page 17, that “Nothing in Jay, Dou, or Glushakov suggests using a graph-based comparison between a training-device qubit topology and candidate devices' qubit topologies to decide which device to use.”
Examiner response:
Applicant’s arguments with respect to claim(s) 1 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument. Examiner further notes that new reference Kong and Hertzberg is used to show obtaining a qubit topology graph of a training device and candidate devices, while Glushakov is used to compare the graphs using graph metrics.
Applicant asserts:
Applicant argues, on page 19-20, that “Claims 5-7, 13-15, and 18 depend from independent claims 1, 9, or 17 respectively. As noted supra, Jay, Dou, and Glushakov do not disclose or suggest each and every element as recited in these independent claims, and Baughman and Hu do not make up for the aforementioned deficiencies of Jay, Dou, and Glushakov.”
Examiner response:
Applicant’s arguments with respect to claim(s) 1 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant asserts:
Applicant argues, on page 20-21, that “Claims 8 and 16 depend from independent claims 1 and 9 respectively. As noted supra, Jay, Dou, Glushakov, Baughman, and Hu do not disclose or suggest each and every element as recited in these independent claims, and Weder fails to make up for the aforementioned deficiencies of Jay, Dou, Glushakov, Baughman, and Hu.”
Examiner response:
Applicant’s arguments with respect to claim(s) 1 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant asserts:
Applicant argues, on page 21-22, that “Claim 19 depends from independent claim 17. As noted supra, Jay, Dou, Glushakov, Baughman, Hu, and Weder do not disclose or suggest each and every element as recited in this independent claim, and Raam fails to make up for the aforementioned deficiencies of Jay, Dou, Glushakov, Baughman, Hu, and Weder.”
Examiner response:
Applicant’s arguments with respect to claim(s) 1 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant asserts:
Applicant argues, on page 22-23, that “Gusat is not citable as prior art reference under AIA 35 U.S.C. 102(b)(2)(C).”
Examiner response:
Applicant’s arguments with respect to claim(s) 1 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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) 1-4 and 9-12 are rejected under 35 U.S.C. 103 as being unpatentable over Jay et al; US 10831455 B2 published Nov 10, 2020 (hereinafter “Jay”) in view of Weicheng Kong; US 20240005190 A1 filed on Sep 29, 2021 (hereinafter “Kong”) in further view of Nathon Wiebe et al; US 20200279185 A1 filed on Feb 28, 2019 (hereinafter “Wiebe”) in further view of Jared Hertzberg et al; US 20200401925 A1 published on Dec 24, 2020 (hereinafter “Hertzberg”) in further view of in further view of Glushakov et al; US 20210349914 A1 filed on Jul 20, 2021 (hereinafter “Glushakov”)
Regarding claim 1, Jay teaches A system, comprising: a memory that stores computer-executable components; and at least one processor that executes at least one of the computer-executable components that (Jay Column 7 Paragraph 4; "data processing system 200 employs a hub architecture… Processing unit 206 may contain one or more processors and may be implemented using one or more heterogeneous processor systems." Examiner notes that a device (data processing system) is operatively coupled to at least one processor);
analyzes metadata of a trained quantum machine learning model submitted by a user for execution for performing a defined task, (Jay Column 9 Paragraph 3; "a user can request a quantum search algorithm in a set of quantum assembly language." Jay Column 10 Paragraph 2; "application 314 identifies a reference to a quantum algorithm in the set of quantum assembly language." Examiner notes that application analyzes (to identify) metadata (set of quantum assembly language) of a trained quantum machine learning model (quantum search algorithm) submitted by a user (a user can request) for execution for performing a defined task (search algorithm));
selects, from a set of quantum computing systems, [using a defined criterion], a quantum computing system for execution of the trained quantum machine learning model, [based on a comparison between the first coupling map of the quantum computing system and respective second coupling maps of the quantum computing systems of the set,] (Jay Column 9 Paragraph 4; "the quantum SDK 320 can select a quantum device from the set of quantum devices 324 to execute the set of quantum assembly language." Examiner notes that device hosting application is selecting a quantum computing system (quantum device) from a set of quantum computing systems (set of quantum devices) for execution of the trained quantum machine learning model (to execute the set of quantum assembly language));
executes, [based on the second coupling map of the selected quantum computing system and] using the [physical] qubits of the selected quantum computing system, the trained quantum machine learning model or an adjusted version of the trained quantum machine learning model to perform the defined task (Jay Column 9 Paragraph 6; "Compiler component 310,330 compiles a quantum circuit from the selected quantum algorithm for execution on the selected quantum device." Examiner notes that device (compiler component) executes using the qubits of the second coupling map of the selected quantum computing system (selected quantum device) the trained quantum machine learning model (selected quantum algorithm) to perform the defined task)
Jay does not teach wherein the metadata comprises a first coupling map generated during training of the trained quantum machine learning model on a first quantum computing system,
and wherein the first coupling map comprises a first connection topology of physical qubits that were employed of the first quantum computing system for the training of the trained quantum machine learning model;
wherein the respective second coupling maps comprise connection topologies of physical qubits of the quantum computing systems of the set
using the physical qubits of the [selected] quantum computing system
However, Kong does teach wherein the metadata comprises a first coupling map generated during [training of the trained quantum machine learning model] on a first quantum computing system, (Kong Paragraph 0005; “enable the complex quantum algorithm to run on the quantum chip in a two-dimensional structure.” Kong Paragraph 0007; “a quantum topology graph optimization method is provided, and the method includes: obtaining a first quantum topology graph of a target quantum algorithm” Examiner notes that metadata is any data that relates to the quantum machine learning model that comprises a first coupling map generated/obtained (first quantum topology graph) during/of a target quantum algorithm on a first quantum computing system (quantum chip))
and wherein the first coupling map comprises a first connection topology of physical qubits that were employed of the first quantum computing system [for the training of the trained quantum machine learning model]; (Kong Paragraph 0005; “enable the complex quantum algorithm to run on the quantum chip in a two-dimensional structure.” Kong Paragraph 0007; “a quantum topology graph optimization method is provided, and the method includes: obtaining a first quantum topology graph of a target quantum algorithm” Kong Paragraph 0078; “Wherein, the qubits may refer to a physical system that can be in a ground state |0>, an excited state |1> and a superposition state (α|0>+β|1>) at the same time.” Examiner notes that a first connection topology of physical qubits (a first quantum topology graph; where qubits refer to a physical system meaning physical qubits) that were employed/enabled of the first quantum computing system to perform quantum algorithm)
using the physical qubits of the [selected] quantum computing system (Kong Paragraph 0349; “improve the execution effect of the quantum algorithm when the target quantum connectivity graph without the crossed connecting lines is implemented on a quantum chip in a two-dimensional structure.” Examiner notes using the physical qubits of the quantum computing system (executing quantum chip which contains physical qubits))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay and Kong. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. One of ordinary skill would have motivation to combine Jay and Kong to improve the execution and fidelity of the quantum algorithm “and in the implementation process, it is possible to reduce the quantity of applied swap gates, and then improve the execution effect of the quantum algorithm; by optimizing the node to be optimized that exceeds the connectivity threshold of the quantum chip to be applied, when the logic bits in the optimized quantum topology graph are mapped to the qubits of the quantum chip, it is possible to reduce the effect of neighboring qubits on the qubit to be regulated through the coupling structure and improve the fidelity of the quantum algorithm” (Kong Paragraph 0013).
Jay in view of Kong does not teach training of the trained quantum machine learning model
for the training of the trained quantum machine learning model
However, Wiebe does teach training of the trained quantum machine learning model (Wiebe Paragraph 0030; “Classical data is traditionally fed to a quantum algorithm in the form of a training set, or test set, of vectors. But rather than train on individual vectors, as one does in classical machine learning, quantum machine learning provides an opportunity to train on quantum-state vectors.” Examiner notes that training of the trained quantum machine learning model (quantum neural network) is a quantum algorithm)
for the training of the trained quantum machine learning model (Examiner refers to previous mapping)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, and Wiebe. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. One of ordinary skill would have motivation to combine Jay, Kong, and Wiebe to overcome various challenges in training a quantum neural network “In the application of quantum computers to neural networks, various challenges persist as to the manner in which a quantum neural network may be trained to a desired task.” (Wiebe Paragraph 0002).
Jay in view of Kong in further view of Wiebe does not teach wherein the respective second coupling maps comprise connection topologies of [physical] qubits of the quantum computing systems of the set
However, Hertzberg does teach wherein the respective second coupling maps comprise connection topologies of [physical] qubits of the quantum computing systems of the set (Hertzberg Paragraph 0032; “simulating operation of qubits in a subgraph topology of a graph representing a topology of a quantum circuit… and/or selecting the quantum circuit topology from a plurality of quantum circuit topologies” Examiner notes that the respective second coupling maps (quantum circuit topology) comprise connection topologies of qubits (qubits in a subgraph topology of a graph) of the quantum computing systems of the set (from a plurality of quantum circuit topologies representing quantum circuits))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, and Hertzberg. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, and Hertzberg to improve the performance of a quantum computer “fewer frequency collisions can improve the performance of a quantum computer by enabling deeper quantum circuits.” (Hertzberg Paragraph 0048).
Jay in view of Kong in further view of Wiebe in further view of Hertzberg does not teach [selects, from a set of quantum computing systems,] using a defined criterion, [a quantum computing system for execution of the trained quantum machine learning model,] based on a comparison between the first coupling map and respective second coupling maps of the quantum computing systems of the set,
based on the second coupling map of the selected quantum computing system and
However, Glushakov does teach [selects, from a set of quantum computing systems,] using a defined criterion, [a quantum computing system for execution of the trained quantum machine learning model,] based on a comparison between the first coupling map and respective second coupling maps of the quantum computing systems of the set, (Glushakov Paragraph 0093; "Thus, each metric graph from the plurality of metric graphs can be compared to its corresponding dull graph. In these embodiments, the Gromov-Hausdorff distances can be calculated between the metric graph and the dull graph corresponding to this metric graph" Examiner notes selection is based on a comparison between the first coupling map (dull graph) and respective second coupling maps of the quantum computing systems of the set (metric graphs) using a defined criterion (Gromov-Hausdorff distances))
based on the second coupling map of the selected quantum computing system and (Examiner refers to previous mapping to show that the selected quantum computing system is selected based comparison of second coupling map)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Regarding claim 2, Jay does not teach The system of claim 1, wherein the first coupling map comprises a first graph whose nodes respectively represent physical qubits of the first quantum computing system and whose edges respectively represent qubit-to-qubit connections of the first quantum computing system,
wherein the respective second coupling map comprise respective second graphs whose nodes respectively represent physical qubits of the quantum computing systems of the set and whose edges respectively represent qubit-to-qubit connections of the quantum computing systems
However, Kong does teach The system of claim 1, wherein the first coupling map comprises a first graph whose nodes respectively represent physical qubits of the first quantum computing system and whose edges respectively represent qubit-to-qubit connections of the first quantum computing system, (Kong Paragraph 0007; “the first quantum topology graph includes a plurality of graph nodes and connecting lines between two graph nodes, wherein the graph nodes are used to represent logic bits in the target quantum algorithm, and the connecting lines are used to represent qubit logic gates between two logic bits;” Kong Paragraph 0164; “Wherein, the quantum connectivity graph can be obtained based on the logic bits in the target quantum algorithm and the number of qubit logic gates applied on any two qubits. Wherein, the qubits may refer to a physical system… Quantum circuits are implemented by manipulating several qubits at the same time.” Examiner notes that first coupling map (quantum topology graph) comprises a first graph whose nodes (graph nodes) respectively represent physical qubits (logic bits/qubits) and whose edges (connecting lines) respectively represent qubit-to-qubit connections (connecting lines are used to represent qubit logic gates between two logic bits) of the first quantum computing system (quantum circuit))
wherein the respective second coupling map comprise respective second graphs whose nodes respectively represent physical qubits of the quantum computing systems of the set and whose edges respectively represent qubit-to-qubit connections of the quantum computing systems (Kong Paragraph 0007; “the first quantum topology graph includes a plurality of graph nodes and connecting lines between two graph nodes, wherein the graph nodes are used to represent logic bits in the target quantum algorithm, and the connecting lines are used to represent qubit logic gates between two logic bits;” Kong Paragraph 0164; “Wherein, the quantum connectivity graph can be obtained based on the logic bits in the target quantum algorithm and the number of qubit logic gates applied on any two qubits. Wherein, the qubits may refer to a physical system… Quantum circuits are implemented by manipulating several qubits at the same time.” Examiner notes that first coupling map (quantum topology graph) comprises a first graph whose nodes (graph nodes) respectively represent physical qubits (logic bits/qubits) and whose edges (connecting lines) respectively represent qubit-to-qubit connections (connecting lines are used to represent qubit logic gates between two logic bits) of the first quantum computing system (quantum circuit))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay and Kong. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. One of ordinary skill would have motivation to combine Jay and Kong to improve the execution and fidelity of the quantum algorithm “and in the implementation process, it is possible to reduce the quantity of applied swap gates, and then improve the execution effect of the quantum algorithm; by optimizing the node to be optimized that exceeds the connectivity threshold of the quantum chip to be applied, when the logic bits in the optimized quantum topology graph are mapped to the qubits of the quantum chip, it is possible to reduce the effect of neighboring qubits on the qubit to be regulated through the coupling structure and improve the fidelity of the quantum algorithm” (Kong Paragraph 0013).
Jay in view of Kong does not teach and wherein the at least one of the computer-executable components further compares the first coupling map to the respective second coupling maps by computing a respective subgraph matching metric.
However, Glushakov does teach and wherein the at least one of the computer-executable components further compares the first coupling map to the respective second coupling maps by computing a respective subgraph matching metric. (Glushakov Paragraph 0093; "Thus, each metric graph from the plurality of metric graphs can be compared to its corresponding dull graph. In these embodiments, the Gromov-Hausdorff distances can be calculated between the metric graph and the dull graph corresponding to this metric graph" Examiner notes a comparison between the first coupling map (dull graph) and respective second coupling map (metric graphs) by computing a respective subgraph matching metric (Gromov-Hausdorff distances))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Regarding claim 3, Jay does not teach The system of claim 2, wherein the respective subgraph matching metric comprises at least one of a Gromov-Hausdorff distance or a Manhattan distance
However, Glushakov does teach The system of claim 2, wherein the respective subgraph matching metric comprises at least one of a Gromov-Hausdorff distance or a Manhattan distance (Glushakov Paragraph 0093; "selecting the optimal graph may include calculating Gromov-Hausdorff distances between each of the metric graphs and a dull graph." Examiner notes that the Gromov-Hausdorff distance is used as the matching metric between the dull graph/first coupling map and metric graph/second coupling map)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Regarding claim 4, Jay teaches and wherein the at least one of the computer-executable components further executes the trained quantum machine learning model on the selected quantum computing system. (Jay Column 9 Paragraph 6; "Compiler component 310,330 compiles a quantum circuit from the selected quantum algorithm for execution on the selected quantum device." Examiner notes that computer executable components (compiler component) executes the trained quantum machine learning model (selected quantum algorithm) on the selected quantum computing system (selected quantum device)
Jay does not teach The system of claim 1, wherein the second coupling map of the selected quantum computing system topologically matches the first coupling map or topologically matches a subgraph of the first coupling map,
However, Glushakov does teaches The system of claim 1, wherein the second coupling map of the selected quantum computing system topologically matches the first coupling map or topologically matches a subgraph of the first coupling map, (Glushakov Paragraph 0091; "In some embodiments, the optimal graph can be determined based on modularity, which is a value that measures how similar a metric graph is to a random graph. The random graph can be constructed using the same set of nodes as the metric graphs in the plurality of metric graphs. A metric graph having the largest value of modularity can be selected as the optimal graph because such metric graph is the least similar to the random graph." Examiner notes that a low or zero value of modularity shows that the graphs/maps are most similar/topologically matching);
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Regarding claim 9, Jay teaches A computer-implemented method, comprising: analyzing, by a device operatively coupled to at least one processor (Jay Column 7 Paragraph 4; "data processing system 200 employs a hub architecture… Processing unit 206 may contain one or more processors and may be implemented using one or more heterogeneous processor systems." Examiner notes that a device (data processing system) is operatively coupled to at least one processor);
analyzing, by a device operatively coupled to at least one processor, metadata of a trained quantum machine learning model submitted by a user for execution for performing a defined task, (Jay Column 9 Paragraph 3; "a user can request a quantum search algorithm in a set of quantum assembly language." Jay Column 10 Paragraph 2; "application 314 identifies a reference to a quantum algorithm in the set of quantum assembly language." Examiner notes that application analyzes (to identify) metadata (set of quantum assembly language) of a trained quantum machine learning model (quantum search algorithm) submitted by a user (a user can request) for execution for performing a defined task (search algorithm));
selecting, by the device and from a set of quantum computing systems, [using a defined criterion], a quantum computing system for execution of the trained quantum machine learning model, [based on a comparison between the first coupling map of the quantum computing system and respective second coupling maps of the quantum computing systems of the set,] (Jay Column 9 Paragraph 4; "the quantum SDK 320 can select a quantum device from the set of quantum devices 324 to execute the set of quantum assembly language." Examiner notes that device hosting application is selecting a quantum computing system (quantum device) from a set of quantum computing systems (set of quantum devices) for execution of the trained quantum machine learning model (to execute the set of quantum assembly language));
and executing, by the device, [based on the second coupling map of the selected quantum computing system and] using the [physical] qubits of the selected quantum computing system, the trained quantum machine learning model or an adjusted version of the trained quantum machine learning model to perform the defined task (Jay Column 9 Paragraph 6; "Compiler component 310,330 compiles a quantum circuit from the selected quantum algorithm for execution on the selected quantum device." Examiner notes that device (compiler component) executes using the qubits of the second coupling map of the selected quantum computing system (selected quantum device) the trained quantum machine learning model (selected quantum algorithm) to perform the defined task)
Jay does not teach wherein the metadata comprises a first coupling map generated during training of the trained quantum machine learning model on a first quantum computing system,
and wherein the first coupling map comprises a first connection topology of physical qubits that were employed of the first quantum computing system for the training of the trained quantum machine learning model;
wherein the respective second coupling maps comprise connection topologies of physical qubits of the quantum computing systems of the set
using the physical qubits of the [selected] quantum computing system
However, Kong does teach wherein the metadata comprises a first coupling map generated during [training of the trained quantum machine learning model] on a first quantum computing system, (Kong Paragraph 0005; “enable the complex quantum algorithm to run on the quantum chip in a two-dimensional structure.” Kong Paragraph 0007; “a quantum topology graph optimization method is provided, and the method includes: obtaining a first quantum topology graph of a target quantum algorithm” Examiner notes that metadata is any data that relates to the quantum machine learning model that comprises a first coupling map generated/obtained (first quantum topology graph) during/of a target quantum algorithm on a first quantum computing system (quantum chip))
and wherein the first coupling map comprises a first connection topology of physical qubits that were employed of the first quantum computing system [for the training of the trained quantum machine learning model]; (Kong Paragraph 0005; “enable the complex quantum algorithm to run on the quantum chip in a two-dimensional structure.” Kong Paragraph 0007; “a quantum topology graph optimization method is provided, and the method includes: obtaining a first quantum topology graph of a target quantum algorithm” Kong Paragraph 0078; “Wherein, the qubits may refer to a physical system that can be in a ground state |0>, an excited state |1> and a superposition state (α|0>+β|1>) at the same time.” Examiner notes that a first connection topology of physical qubits (a first quantum topology graph; where qubits refer to a physical system meaning physical qubits) that were employed/enabled of the first quantum computing system to perform quantum algorithm)
using the physical qubits of the [selected] quantum computing system (Kong Paragraph 0349; “improve the execution effect of the quantum algorithm when the target quantum connectivity graph without the crossed connecting lines is implemented on a quantum chip in a two-dimensional structure.” Examiner notes using the physical qubits of the quantum computing system (executing quantum chip which contains physical qubits))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay and Kong. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. One of ordinary skill would have motivation to combine Jay and Kong to improve the execution and fidelity of the quantum algorithm “and in the implementation process, it is possible to reduce the quantity of applied swap gates, and then improve the execution effect of the quantum algorithm; by optimizing the node to be optimized that exceeds the connectivity threshold of the quantum chip to be applied, when the logic bits in the optimized quantum topology graph are mapped to the qubits of the quantum chip, it is possible to reduce the effect of neighboring qubits on the qubit to be regulated through the coupling structure and improve the fidelity of the quantum algorithm” (Kong Paragraph 0013).
Jay in view of Kong does not teach training of the trained quantum machine learning model
for the training of the trained quantum machine learning model
However, Wiebe does teach training of the trained quantum machine learning model (Wiebe Paragraph 0030; “Classical data is traditionally fed to a quantum algorithm in the form of a training set, or test set, of vectors. But rather than train on individual vectors, as one does in classical machine learning, quantum machine learning provides an opportunity to train on quantum-state vectors.” Examiner notes that training of the trained quantum machine learning model (quantum neural network) is a quantum algorithm)
for the training of the trained quantum machine learning model (Examiner refers to previous mapping)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, and Wiebe. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. One of ordinary skill would have motivation to combine Jay, Kong, and Wiebe to overcome various challenges in training a quantum neural network “In the application of quantum computers to neural networks, various challenges persist as to the manner in which a quantum neural network may be trained to a desired task.” (Wiebe Paragraph 0002).
Jay in view of Kong in further view of Wiebe does not teach wherein the respective second coupling maps comprise connection topologies of [physical] qubits of the quantum computing systems of the set
However, Hertzberg does teach wherein the respective second coupling maps comprise connection topologies of [physical] qubits of the quantum computing systems of the set (Hertzberg Paragraph 0032; “simulating operation of qubits in a subgraph topology of a graph representing a topology of a quantum circuit… and/or selecting the quantum circuit topology from a plurality of quantum circuit topologies” Examiner notes that the respective second coupling maps (quantum circuit topology) comprise connection topologies of qubits (qubits in a subgraph topology of a graph) of the quantum computing systems of the set (from a plurality of quantum circuit topologies representing quantum circuits))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, and Hertzberg. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, and Hertzberg to improve the performance of a quantum computer “fewer frequency collisions can improve the performance of a quantum computer by enabling deeper quantum circuits.” (Hertzberg Paragraph 0048).
Jay in view of Kong in further view of Wiebe in further view of Hertzberg does not teach [Selecting, by the device and from a set of quantum computing systems] using a defined criterion, [a quantum computing system for execution of the trained quantum machine learning model,] based on a comparison between the first coupling map and respective second coupling maps of the quantum computing systems of the set,
based on the second coupling map of the selected quantum computing system and
However, Glushakov does teach [Selecting, by the device and from a set of quantum computing systems] using a defined criterion, [a quantum computing system for execution of the trained quantum machine learning model,] based on a comparison between the first coupling map and respective second coupling maps of the quantum computing systems of the set, (Glushakov Paragraph 0093; "Thus, each metric graph from the plurality of metric graphs can be compared to its corresponding dull graph. In these embodiments, the Gromov-Hausdorff distances can be calculated between the metric graph and the dull graph corresponding to this metric graph" Examiner notes selection is based on a comparison between the first coupling map (dull graph) and respective second coupling maps of the quantum computing systems of the set (metric graphs) using a defined criterion (Gromov-Hausdorff distances))
based on the second coupling map of the selected quantum computing system and (Examiner refers to previous mapping to show that the selected quantum computing system is selected based comparison of second coupling map)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Regarding claim 10, Jay does not teach The computer-implemented method of claim 9, wherein the first coupling map comprises a first graph whose nodes respectively represent physical qubits of the first quantum computing system and whose edges respectively represent qubit-to-qubit connections of the first quantum computing system,
wherein the respective second coupling map comprise respective second graphs whose nodes respectively represent physical qubits of the quantum computing systems of the set and whose edges respectively represent qubit-to-qubit connections of the quantum computing systems
However, Kong does teach The computer-implemented method of claim 9, wherein the first coupling map comprises a first graph whose nodes respectively represent physical qubits of the first quantum computing system and whose edges respectively represent qubit-to-qubit connections of the first quantum computing system, (Kong Paragraph 0007; “the first quantum topology graph includes a plurality of graph nodes and connecting lines between two graph nodes, wherein the graph nodes are used to represent logic bits in the target quantum algorithm, and the connecting lines are used to represent qubit logic gates between two logic bits;” Kong Paragraph 0164; “Wherein, the quantum connectivity graph can be obtained based on the logic bits in the target quantum algorithm and the number of qubit logic gates applied on any two qubits. Wherein, the qubits may refer to a physical system… Quantum circuits are implemented by manipulating several qubits at the same time.” Examiner notes that first coupling map (quantum topology graph) comprises a first graph whose nodes (graph nodes) respectively represent physical qubits (logic bits/qubits) and whose edges (connecting lines) respectively represent qubit-to-qubit connections (connecting lines are used to represent qubit logic gates between two logic bits) of the first quantum computing system (quantum circuit))
wherein the respective second coupling map comprise respective second graphs whose nodes respectively represent physical qubits of the quantum computing systems of the set and whose edges respectively represent qubit-to-qubit connections of the quantum computing systems (Kong Paragraph 0007; “the first quantum topology graph includes a plurality of graph nodes and connecting lines between two graph nodes, wherein the graph nodes are used to represent logic bits in the target quantum algorithm, and the connecting lines are used to represent qubit logic gates between two logic bits;” Kong Paragraph 0164; “Wherein, the quantum connectivity graph can be obtained based on the logic bits in the target quantum algorithm and the number of qubit logic gates applied on any two qubits. Wherein, the qubits may refer to a physical system… Quantum circuits are implemented by manipulating several qubits at the same time.” Examiner notes that first coupling map (quantum topology graph) comprises a first graph whose nodes (graph nodes) respectively represent physical qubits (logic bits/qubits) and whose edges (connecting lines) respectively represent qubit-to-qubit connections (connecting lines are used to represent qubit logic gates between two logic bits) of the first quantum computing system (quantum circuit))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay and Kong. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. One of ordinary skill would have motivation to combine Jay and Kong to improve the execution and fidelity of the quantum algorithm “and in the implementation process, it is possible to reduce the quantity of applied swap gates, and then improve the execution effect of the quantum algorithm; by optimizing the node to be optimized that exceeds the connectivity threshold of the quantum chip to be applied, when the logic bits in the optimized quantum topology graph are mapped to the qubits of the quantum chip, it is possible to reduce the effect of neighboring qubits on the qubit to be regulated through the coupling structure and improve the fidelity of the quantum algorithm” (Kong Paragraph 0013).
Jay in view of Kong does not teach and further comprising: facilitating, by the device, the comparison between the first coupling map and the respective second coupling maps by computing a respective subgraph matching metric.
However, Glushakov does teach and further comprising: facilitating, by the device, the comparison between the first coupling map and the respective second coupling maps by computing a respective subgraph matching metric. (Glushakov Paragraph 0093; "Thus, each metric graph from the plurality of metric graphs can be compared to its corresponding dull graph. In these embodiments, the Gromov-Hausdorff distances can be calculated between the metric graph and the dull graph corresponding to this metric graph" Examiner notes a comparison between the first coupling map (dull graph) and respective second coupling map (metric graphs) by computing a respective subgraph matching metric (Gromov-Hausdorff distances))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Regarding claim 11, Jay does not teach The computer-implemented method of claim 10, wherein the respective subgraph matching metric comprises at least one of a Gromov-Hausdorff distance or a Manhattan distance
However, Glushakov does teach The computer-implemented method of claim 10, wherein the respective subgraph matching metric comprises at least one of a Gromov-Hausdorff distance or a Manhattan distance (Glushakov Paragraph 0093; "selecting the optimal graph may include calculating Gromov-Hausdorff distances between each of the metric graphs and a dull graph." Examiner notes that the Gromov-Hausdorff distance is used as the matching metric between the dull graph/first coupling map and metric graph/second coupling map)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Regarding claim 12, Jay teaches and wherein the device is configured to execute the trained quantum machine learning model on the selected quantum computing system. (Jay Column 9 Paragraph 6; "Compiler component 310,330 compiles a quantum circuit from the selected quantum algorithm for execution on the selected quantum device." Examiner notes that device (compiler component) executes the trained quantum machine learning model (selected quantum algorithm) on the selected quantum computing system (selected quantum device)
Jay does not teach The computer-implemented method of claim 9, wherein the second coupling map of the selected quantum computing system topologically matches the first coupling map or topologically matches a subgraph of the first coupling map,
However, Glushakov does teaches The computer-implemented method of claim 9, wherein the second coupling map of the selected quantum computing system topologically matches the first coupling map or topologically matches a subgraph of the first coupling map, (Glushakov Paragraph 0091; "In some embodiments, the optimal graph can be determined based on modularity, which is a value that measures how similar a metric graph is to a random graph. The random graph can be constructed using the same set of nodes as the metric graphs in the plurality of metric graphs. A metric graph having the largest value of modularity can be selected as the optimal graph because such metric graph is the least similar to the random graph." Examiner notes that a low or zero value of modularity shows that the graphs/maps are most similar/topologically matching);
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Claim(s) 5-6, 13-14, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Jay et al; US 10831455 B2 published Nov 10, 2020 (hereinafter “Jay”) in view of Weicheng Kong; US 20240005190 A1 filed on Sep 29, 2021 (hereinafter “Kong”) in further view of Nathon Wiebe et al; US 20200279185 A1filed on Feb 28, 2019 (hereinafter “Wiebe”) in further view of Jared Hertzberg et al; US 20200401925 A1 published on Dec 24, 2020 (hereinafter “Hertzberg”) in further view of in further view of Glushakov et al; US 20210349914 A1 filed on Jul 20, 2021 (hereinafter “Glushakov”) in further view of Aaron Baughman; “Study of Feature Importance for Quantum Machine Learning Models” available Jun 9, 2022 (hereinafter “Aaron”) in further view of Zhirui et al; “Quantum Neural Network Compression” available Jul 5, 2022 (hereinafter “Zhirui”)
Regarding claim 5, Jay teaches and executes the adjusted version of the trained quantum machine learning model on the selected quantum computing system. (Jay Column 9 Paragraph 6; "Compiler component 310,330 compiles a quantum circuit from the selected quantum algorithm for execution on the selected quantum device." Examiner notes that device (compiler component) executes the adjusted version of the trained quantum computing system (selected quantum algorithm on the selected quantum computing system (selected quantum device))
Jay does not teach The system of claim 2, wherein the second coupling map of the selected quantum computing system does not topologically match the first coupling map and does not topologically match any subgraph of the first coupling map,
However, Glushakov does teach The system of claim 2, wherein the second coupling map of the selected quantum computing system does not topologically match the first coupling map and does not topologically match any subgraph of the first coupling map, (Glushakov Paragraph 0093; "A metric graph having the largest distance from the dull graph can be selected." Examiner notes that the largest Gromov-Hausdorff distance indicates the graphs/maps do not topologically match);
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Jay in view of Glushakov does not teach based on the set of variable importance scores;
and wherein the at least one of the computer-executable components further: accesses a set of variable importance scores associated with the trained quantum machine learning model;
However, Aaron does teach adjusts, based on the set of variable importance scores, the quantum machine learning model; (Aaron Section 3.A Paragraph 1; "This algorithm can be applied with any opaque estimators represented by a fitted predictive model to compute the reference scores of the model on data set [10]. We extend this technique to calculate the feature importance of quantum models with the following steps:" Examiner notes that Section 3.A talks about an algorithm to calculate feature importance and how to apply it to a quantum machine learning model shows adjusting the quantum computing system based on variable importance scores)
and wherein the at least one of the computer-executable components further: accesses a set of variable importance scores associated with the trained quantum machine learning model;
(Aaron Section 3; "Throughout this work, we applied two different types of feature importance algorithms: 1. Permutation Importance [7] 2. ALE Accumulated Local Effects (ALE) for feature importance [8]" Examiner notes that Section 3 talks about ways to get/access feature importance/variable importance scores; Aaron Section 4 Table 1 shows computing device with computer executable components to perform action);
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, and Aaron. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, and Aaron to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “Feature importance scores play an important role in a predictive modeling project such as providing insight into the data and mode” (Aaron Section 2 Paragraph 1).
Jay in view of Glushakov in further view of Aaron does not teach adjusts, [based on the set of variable importance scores], the trained quantum machine learning model, thereby yielding the adjusted version of the trained quantum machine learning model;
However, Zhirui does teach adjusts, [based on the set of variable importance scores], the trained quantum machine learning model, thereby yielding the adjusted version of the trained quantum machine learning model; (Zhirui Section 3 Subsection “QNN Pruning” mentions adjusting the quantum machine learning model (Quantum Neural Network) by pruning, thereby yielding the adjusted version of the quantum machine learning model (pruned QNN));
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui to reduce the model size and speed up its execution “Deep Neural Network (DNN) Compression. Pruning [2, 6, 12] and quantization [17, 18, 21] are effective in reducing the model size and speed up its execution” (Aaron Section 2 Paragraph 1).
Regarding claim 6, Jay does not teach The system of claim 5, wherein the at least one of the computer-executable components further: determines based on one or more executions of the adjusted version of the trained quantum machine learning model, an accuracy level of the adjusted version of the trained quantum machine learning model.
However, Zhirui does teach The computer-implemented method of claim 13, further comprising: determining, by the device and via one or more executions of the adjusted version of the trained quantum machine learning model, an accuracy level of the adjusted version of the trained quantum machine learning model. (Zhirui Section 4.1 Paragraph 4 and Figure 6; "We evaluate CompVQC on 2 common classification datasets… For all datasets, we apply 90% samples for the train set and 10% for test set." Examiner notes that an accuracy level (accuracy percentage from Fig 6) of the adjusted version of the trained quantum machine learning model (adjusted QNN from CompVQC framework) is determined by the computer executable components and via one or more executions of the adjusted version of the trained quantum machine learning model (executing CompVQC on test set))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui to reduce the model size and speed up its execution “Deep Neural Network (DNN) Compression. Pruning [2, 6, 12] and quantization [17, 18, 21] are effective in reducing the model size and speed up its execution” (Aaron Section 2 Paragraph 1).
Regarding claim 13, Jay teaches wherein the device is configured to execute the adjusted version of the trained quantum machine learning model on the selected quantum computing system. (Jay Column 9 Paragraph 6; "Compiler component 310,330 compiles a quantum circuit from the selected quantum algorithm for execution on the selected quantum device." Examiner notes that device (compiler component) executes the adjusted version of the trained quantum computing system (selected quantum algorithm on the selected quantum computing system (selected quantum device))
Jay does not teach The computer-implemented method of claim 10, wherein the second coupling map of the selected quantum computing system does not topologically match the first coupling map and does not topologically match any subgraph of the first coupling map,
However, Glushakov does teach The computer-implemented method of claim 10, wherein the second coupling map of the selected quantum computing system does not topologically match the first coupling map and does not topologically match any subgraph of the first coupling map, (Glushakov Paragraph 0093; "A metric graph having the largest distance from the dull graph can be selected." Examiner notes that the largest Gromov-Hausdorff distance indicates the graphs/maps do not topologically match);
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Jay in view of Glushakov does not teach by the device and based on the set of variable importance scores;
and further comprising: accessing, by the device, a set of variable importance scores associated with the quantum machine learning model;
However, Aaron does teach adjusting, by the device and based on the set of variable importance scores, the quantum machine learning model; (Aaron Section 3.A Paragraph 1; "This algorithm can be applied with any opaque estimators represented by a fitted predictive model to compute the reference scores of the model on data set [10]. We extend this technique to calculate the feature importance of quantum models with the following steps:" Examiner notes that Section 3.A talks about an algorithm to calculate feature importance and how to apply it to a quantum machine learning model shows adjusting the quantum computing system based on variable importance scores)
and further comprising: accessing, by the device, a set of variable importance scores associated with the quantum machine learning model;
(Aaron Section 3; "Throughout this work, we applied two different types of feature importance algorithms: 1. Permutation Importance [7] 2. ALE Accumulated Local Effects (ALE) for feature importance [8]" Examiner notes that Section 3 talks about ways to get/access feature importance/variable importance scores; Aaron Section 4 Table 1 shows computing device to perform action);
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, and Aaron. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, and Aaron to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “Feature importance scores play an important role in a predictive modeling project such as providing insight into the data and mode” (Aaron Section 2 Paragraph 1).
Jay in view of Glushakov in further view of Aaron does not teach and adjusting, [by the device and based on the set of variable importance scores], the quantum machine learning model, thereby yielding the adjusted version of the quantum machine learning model, wherein the device is configured to execute the adjusted version of the quantum machine learning model on the quantum computing system.
However, Zhirui does teach and adjusting, [by the device and based on the set of variable importance scores], the quantum machine learning model, thereby yielding the adjusted version of the quantum machine learning model, (Zhirui Section 3 Subsection “QNN Pruning” mentions adjusting the quantum machine learning model (Quantum Neural Network) by pruning, thereby yielding the adjusted version of the quantum machine learning model (pruned QNN));
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui to reduce the model size and speed up its execution “Deep Neural Network (DNN) Compression. Pruning [2, 6, 12] and quantization [17, 18, 21] are effective in reducing the model size and speed up its execution” (Aaron Section 2 Paragraph 1).
Regarding claim 14, Jay does not teach The computer-implemented method of claim 13, further comprising: determining, by the device and via one or more executions of the adjusted version of the trained quantum machine learning model, an accuracy level of the adjusted version of the trained quantum machine learning model.
However, Zhirui does teach The computer-implemented method of claim 13, further comprising: determining, by the device and via one or more executions of the adjusted version of the trained quantum machine learning model, an accuracy level of the adjusted version of the trained quantum machine learning model. (Zhirui Section 4.1 Paragraph 4 and Figure 6; "We evaluate CompVQC on 2 common classification datasets… For all datasets, we apply 90% samples for the train set and 10% for test set." Examiner notes that an accuracy level (accuracy percentage from Fig 6) of the adjusted version of the trained quantum machine learning model (adjusted QNN from CompVQC framework) is determined by the device and via one or more executions of the adjusted version of the trained quantum machine learning model (executing CompVQC on test set))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui to reduce the model size and speed up its execution “Deep Neural Network (DNN) Compression. Pruning [2, 6, 12] and quantization [17, 18, 21] are effective in reducing the model size and speed up its execution” (Aaron Section 2 Paragraph 1).
Regarding claim 17, Jay teaches A computer program product for facilitating quantum system selection via coupling map comparison, the computer program product comprising at least one non-transitory computer-readable memory having program instructions embodied therewith, the program instructions executable by at least one processor to cause the at least one processor to: (Jay Column 7 Paragraph 4; "data processing system 200 employs a hub architecture… Processing unit 206 may contain one or more processors and may be implemented using one or more heterogeneous processor systems." Examiner notes that a device (data processing system) is operatively coupled to at least one processor);
analyze metadata of a trained quantum machine learning model submitted by a user for execution for performing a defined task (Jay Column 9 Paragraph 3; "a user can request a quantum search algorithm in a set of quantum assembly language." Jay Column 10 Paragraph 2; "application 314 identifies a reference to a quantum algorithm in the set of quantum assembly language." Examiner notes that application analyzes (to identify) metadata (set of quantum assembly language) of a trained quantum machine learning model (quantum search algorithm) submitted by a user (a user can request) for execution for performing a defined task (search algorithm));
select, from a set of quantum computing systems, [using a defined criterion], a quantum computing system for execution of the trained quantum machine learning model, [based on a comparison between the first coupling map and respective second coupling maps of the quantum computing systems of the set, (Jay Column 9 Paragraph 4; "the quantum SDK 320 can select a quantum device from the set of quantum devices 324 to execute the set of quantum assembly language." Examiner notes that device hosting application is selecting a quantum computing system (quantum device) from a set of quantum computing systems (set of quantum devices) for execution of the trained quantum machine learning model (to execute the set of quantum assembly language));
executes, [based on the second coupling map of the selected quantum computing system and] using the [physical] qubits of the selected quantum computing system, the trained quantum machine learning model or an adjusted version of the trained quantum machine learning model to perform the defined task (Jay Column 9 Paragraph 6; "Compiler component 310,330 compiles a quantum circuit from the selected quantum algorithm for execution on the selected quantum device." Examiner notes that device (compiler component) executes using the qubits of the second coupling map of the selected quantum computing system (selected quantum device) the trained quantum machine learning model (selected quantum algorithm) to perform the defined task)
Jay does not teach wherein the metadata comprises a first coupling map generated during training of the trained quantum machine learning model on a first quantum computing system,
and wherein the first coupling map comprises a first connection topology of physical qubits that were employed of the first quantum computing system for the training of the trained quantum machine learning model;
wherein the respective second coupling maps comprise connection topologies of physical qubits of the quantum computing systems of the set
using the physical qubits of the [selected] quantum computing system
However, Kong does teach wherein the metadata comprises a first coupling map generated during [training of the trained quantum machine learning model] on a first quantum computing system, (Kong Paragraph 0005; “enable the complex quantum algorithm to run on the quantum chip in a two-dimensional structure.” Kong Paragraph 0007; “a quantum topology graph optimization method is provided, and the method includes: obtaining a first quantum topology graph of a target quantum algorithm” Examiner notes that metadata is any data that relates to the quantum machine learning model that comprises a first coupling map generated/obtained (first quantum topology graph) during/of a target quantum algorithm on a first quantum computing system (quantum chip))
and wherein the first coupling map comprises a first connection topology of physical qubits that were employed of the first quantum computing system [for the training of the trained quantum machine learning model]; (Kong Paragraph 0005; “enable the complex quantum algorithm to run on the quantum chip in a two-dimensional structure.” Kong Paragraph 0007; “a quantum topology graph optimization method is provided, and the method includes: obtaining a first quantum topology graph of a target quantum algorithm” Kong Paragraph 0078; “Wherein, the qubits may refer to a physical system that can be in a ground state |0>, an excited state |1> and a superposition state (α|0>+β|1>) at the same time.” Examiner notes that a first connection topology of physical qubits (a first quantum topology graph; where qubits refer to a physical system meaning physical qubits) that were employed/enabled of the first quantum computing system to perform quantum algorithm)
using the physical qubits of the [selected] quantum computing system (Kong Paragraph 0349; “improve the execution effect of the quantum algorithm when the target quantum connectivity graph without the crossed connecting lines is implemented on a quantum chip in a two-dimensional structure.” Examiner notes using the physical qubits of the quantum computing system (executing quantum chip which contains physical qubits))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay and Kong. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. One of ordinary skill would have motivation to combine Jay and Kong to improve the execution and fidelity of the quantum algorithm “and in the implementation process, it is possible to reduce the quantity of applied swap gates, and then improve the execution effect of the quantum algorithm; by optimizing the node to be optimized that exceeds the connectivity threshold of the quantum chip to be applied, when the logic bits in the optimized quantum topology graph are mapped to the qubits of the quantum chip, it is possible to reduce the effect of neighboring qubits on the qubit to be regulated through the coupling structure and improve the fidelity of the quantum algorithm” (Kong Paragraph 0013).
Jay in view of Kong does not teach training of the trained quantum machine learning model
for the training of the trained quantum machine learning model
However, Wiebe does teach training of the trained quantum machine learning model (Wiebe Paragraph 0030; “Classical data is traditionally fed to a quantum algorithm in the form of a training set, or test set, of vectors. But rather than train on individual vectors, as one does in classical machine learning, quantum machine learning provides an opportunity to train on quantum-state vectors.” Examiner notes that training of the trained quantum machine learning model (quantum neural network) is a quantum algorithm)
for the training of the trained quantum machine learning model (Examiner refers to previous mapping)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, and Wiebe. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. One of ordinary skill would have motivation to combine Jay, Kong, and Wiebe to overcome various challenges in training a quantum neural network “In the application of quantum computers to neural networks, various challenges persist as to the manner in which a quantum neural network may be trained to a desired task.” (Wiebe Paragraph 0002).
Jay in view of Kong in further view of Wiebe does not teach wherein the respective second coupling maps comprise connection topologies of [physical] qubits of the quantum computing systems of the set
However, Hertzberg does teach wherein the respective second coupling maps comprise connection topologies of [physical] qubits of the quantum computing systems of the set (Hertzberg Paragraph 0032; “simulating operation of qubits in a subgraph topology of a graph representing a topology of a quantum circuit… and/or selecting the quantum circuit topology from a plurality of quantum circuit topologies” Examiner notes that the respective second coupling maps (quantum circuit topology) comprise connection topologies of qubits (qubits in a subgraph topology of a graph) of the quantum computing systems of the set (from a plurality of quantum circuit topologies representing quantum circuits))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, and Hertzberg. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, and Hertzberg to improve the performance of a quantum computer “fewer frequency collisions can improve the performance of a quantum computer by enabling deeper quantum circuits.” (Hertzberg Paragraph 0048).
Jay in view of Kong in further view of Wiebe in further view of Hertzberg does not teach [selects, from a set of quantum computing systems,] using a defined criterion, [a quantum computing system for execution of the trained quantum machine learning model,] based on a comparison between the first coupling map and respective second coupling maps of the quantum computing systems of the set,
based on the second coupling map of the selected quantum computing system and
However, Glushakov does teach [selects, from a set of quantum computing systems,] using a defined criterion, [a quantum computing system for execution of the trained quantum machine learning model,] based on a comparison between the first coupling map and respective second coupling maps of the quantum computing systems of the set, (Glushakov Paragraph 0093; "Thus, each metric graph from the plurality of metric graphs can be compared to its corresponding dull graph. In these embodiments, the Gromov-Hausdorff distances can be calculated between the metric graph and the dull graph corresponding to this metric graph" Examiner notes selection is based on a comparison between the first coupling map (dull graph) and respective second coupling maps of the quantum computing systems of the set (metric graphs) using a defined criterion (Gromov-Hausdorff distances))
based on the second coupling map of the selected quantum computing system and (Examiner refers to previous mapping to show that the selected quantum computing system is selected based comparison of second coupling map)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, and Glushakov to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “the optimal graph can be determined using the Gromov-Hausdorff distance.” (Glushakov Paragraph 0093).
Jay in view of Kong in further view of Wiebe in further view of Hertzberg in further view of Glushakov does not teach and a set of variable importance scores associated with the trained quantum machine learning model,
adjust, based on the set of variable importance scores, [the trained quantum machine learning model, thereby yielding an adjusted version of the trained quantum machine learning model;]
However, Aaron does teach and a set of variable importance scores associated with the trained quantum machine learning model, Aaron Section 3; "Throughout this work, we applied two different types of feature importance algorithms: 1. Permutation Importance [7] 2. ALE Accumulated Local Effects (ALE) for feature importance [8]" Examiner notes that Section 3 talks about ways to get/access feature importance/variable importance scores associated with the trained quantum machine learning model; Aaron Section 4 Table 1 shows computing device to perform action
based on the set of variable importance scores; (Aaron Section 3.A Paragraph 1; "This algorithm can be applied with any opaque estimators represented by a fitted predictive model to compute the reference scores of the model on data set [10]. We extend this technique to calculate the feature importance of quantum models with the following steps:" Examiner notes that Section 3.A talks about an algorithm to calculate feature importance and how to apply it to a quantum machine learning model shows adjusting the quantum computing system based on variable importance scores)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, and Aaron. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, and Aaron to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “Feature importance scores play an important role in a predictive modeling project such as providing insight into the data and mode” (Aaron Section 2 Paragraph 1).
Jay in view of Kong in further view of Wiebe in further view of Hertzberg in further view of Glushakov in further view of Aaron does not teach adjust, based on the set of variable importance scores, the trained quantum machine learning model, thereby yielding an adjusted version of the trained quantum machine learning model;
However, Zhirui does teach adjust, [based on the set of variable importance scores], the trained quantum machine learning model, thereby yielding an adjusted version of the trained quantum machine learning model; (Zhirui Section 3 Subsection “QNN Pruning” mentions adjusting the quantum machine learning model (Quantum Neural Network) by pruning, thereby yielding the adjusted version of the quantum machine learning model (pruned QNN))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui to reduce the model size and speed up its execution “Deep Neural Network (DNN) Compression. Pruning [2, 6, 12] and quantization [17, 18, 21] are effective in reducing the model size and speed up its execution” (Aaron Section 2 Paragraph 1).
Claim(s) 7, 15, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Jay et al; US 10831455 B2 published Nov 10, 2020 (hereinafter “Jay”) in view of Weicheng Kong; US 20240005190 A1 filed on Sep 29, 2021 (hereinafter “Kong”) in further view of Nathon Wiebe et al; US 20200279185 A1filed on Feb 28, 2019 (hereinafter “Wiebe”) in further view of Jared Hertzberg et al; US 20200401925 A1 published on Dec 24, 2020 (hereinafter “Hertzberg”) in further view of in further view of Glushakov et al; US 20210349914 A1 filed on Jul 20, 2021 (hereinafter “Glushakov”) in further view of Aaron Baughman; “Study of Feature Importance for Quantum Machine Learning Models” available Jun 9, 2022 (hereinafter “Aaron”) in further view of Zhirui et al; “Quantum Neural Network Compression” available Jul 5, 2022 (hereinafter “Zhirui”) in further view of Dou et al; US 20240160977 A1 filed on Mar 30, 2022 (hereinafter “Dou”)
Regarding claim 7, Jay does not teach The system of claim 5, wherein the at least one of the computer-executable components further adjusts the trained quantum machine learning model by iteratively removing, in order of increasing variable importance score, at least one logical qubit or at least one quantum gate from the trained quantum machine learning model.
However, Zhirui does teach The system of claim 5, wherein the at least one of the computer-executable components further adjusts the trained quantum machine learning model by iteratively removing, [in order of increasing variable importance score], at least one logical qubit or at least one quantum gate from the trained quantum machine learning model. (Zhirui Section 3 Paragraph 6; "Lemma 3.1 Given a quantum gate G and parameter x, if the function of G(x) is to multiply the identical matrix or the negative of identical matrix, then G(x) can be pruned, and x is thus a pruning level for G." Zhirui Section 4.2 Paragraph 2; "It is not difficult to understand this result that pruning removes entire gates" Examiner notes that adjusting the trained quantum machine learning model (Quantum Neural Network) includes iteratively removing at least one quantum gate from the trained quantum machine learning model (the plurality of gates is given to lemma 3.1 to be iteratively pruned))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui to reduce the model size and speed up its execution “Deep Neural Network (DNN) Compression. Pruning [2, 6, 12] and quantization [17, 18, 21] are effective in reducing the model size and speed up its execution” (Aaron Section 2 Paragraph 1).
Jay in view of Zhirui does not teach in order of increasing variable importance score;
However, Dou does teach in order of increasing variable importance score; (Dou Section 5.C Paragraph 4;"1 ≤ rank ascending orderi ≤ 146; where rank ascending orderi ∈ Z Ranking of feature importance score is done using ascending order is considered" Examiner notes that feature importance scores/variable importance scores are considered in ascending/increasing order)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Dou. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. Dou teaches determining a topological structure of a quantum circuit. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Dou to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “This not only improves the development efficiency of quantum compilers, but also achieves the technical effect of adapting quantum circuits to arbitrary quantum chip instruction sets.” (Dou Paragraph 0018).
Regarding claim 15, Jay does not teach The computer-implemented method of claim 13, wherein the adjusting the trained quantum machine learning model includes iteratively removing, [by the device and in order of increasing variable importance score,] at least one logical qubit or at least one quantum gate from the trained quantum machine learning model.
However, Zhirui does teach The computer-implemented method of claim 13, wherein the adjusting the trained quantum machine learning model includes iteratively removing, by the device and [in order of increasing variable importance score,] at least one logical qubit or at least one quantum gate from the trained quantum machine learning model. (Zhirui Section 3 Paragraph 6; "Lemma 3.1 Given a quantum gate G and parameter x, if the function of G(x) is to multiply the identical matrix or the negative of identical matrix, then G(x) can be pruned, and x is thus a pruning level for G." Zhirui Section 4.2 Paragraph 2; "It is not difficult to understand this result that pruning removes entire gates" Examiner notes that adjusting the trained quantum machine learning model (Quantum Neural Network) includes iteratively removing at least one quantum gate from the trained quantum machine learning model (the plurality of gates is given to lemma 3.1 to be iteratively pruned))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui to reduce the model size and speed up its execution “Deep Neural Network (DNN) Compression. Pruning [2, 6, 12] and quantization [17, 18, 21] are effective in reducing the model size and speed up its execution” (Aaron Section 2 Paragraph 1).
Jay in view of Zhirui does not teach in order of increasing variable importance score;
However, Dou does teach in order of increasing variable importance score; (Dou Section 5.C Paragraph 4;"1 ≤ rank ascending orderi ≤ 146; where rank ascending orderi ∈ Z Ranking of feature importance score is done using ascending order is considered" Examiner notes that feature importance scores/variable importance scores are considered in ascending/increasing order)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Dou. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. Dou teaches determining a topological structure of a quantum circuit. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Dou to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “This not only improves the development efficiency of quantum compilers, but also achieves the technical effect of adapting quantum circuits to arbitrary quantum chip instruction sets.” (Dou Paragraph 0018).
Regarding claim 18, Jay does not teach The computer program product of claim 17, wherein the at least one processor is configured to adjust the trained quantum machine learning model by iteratively removing, [in order of increasing variable importance score], at least one logical qubit or at least one quantum gate from the trained quantum machine learning model.
However, Zhirui does teach The computer program product of claim 17, wherein the at least one processor is configured to adjust the trained quantum machine learning model by iteratively removing, [in order of increasing variable importance score], at least one logical qubit or at least one quantum gate from the trained quantum machine learning model. (Zhirui Section 3 Paragraph 6; "Lemma 3.1 Given a quantum gate G and parameter x, if the function of G(x) is to multiply the identical matrix or the negative of identical matrix, then G(x) can be pruned, and x is thus a pruning level for G." Zhirui Section 4.2 Paragraph 2; "It is not difficult to understand this result that pruning removes entire gates" Examiner notes that adjusting the trained quantum machine learning model (Quantum Neural Network) includes iteratively removing at least one quantum gate from the trained quantum machine learning model (the plurality of gates is given to lemma 3.1 to be iteratively pruned))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, and Zhirui to reduce the model size and speed up its execution “Deep Neural Network (DNN) Compression. Pruning [2, 6, 12] and quantization [17, 18, 21] are effective in reducing the model size and speed up its execution” (Aaron Section 2 Paragraph 1).
Jay in view of Zhirui does not teach in order of increasing variable importance score;
However, Dou does teach in order of increasing variable importance score; (Dou Section 5.C Paragraph 4;"1 ≤ rank ascending orderi ≤ 146; where rank ascending orderi ∈ Z Ranking of feature importance score is done using ascending order is considered" Examiner notes that feature importance scores/variable importance scores are considered in ascending/increasing order)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Dou. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. Dou teaches determining a topological structure of a quantum circuit. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Dou to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps “This not only improves the development efficiency of quantum compilers, but also achieves the technical effect of adapting quantum circuits to arbitrary quantum chip instruction sets.” (Dou Paragraph 0018).
Claim(s) 8 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Jay et al; US 10831455 B2 published Nov 10, 2020 (hereinafter “Jay”) in view of Weicheng Kong; US 20240005190 A1 filed on Sep 29, 2021 (hereinafter “Kong”) in further view of Nathon Wiebe et al; US 20200279185 A1filed on Feb 28, 2019 (hereinafter “Wiebe”) in further view of Jared Hertzberg et al; US 20200401925 A1 published on Dec 24, 2020 (hereinafter “Hertzberg”) in further view of in further view of Glushakov et al; US 20210349914 A1 filed on Jul 20, 2021 (hereinafter “Glushakov”) in further view of Aaron Baughman; “Study of Feature Importance for Quantum Machine Learning Models” available Jun 9, 2022 (hereinafter “Aaron”) in further view of Zhirui et al; “Quantum Neural Network Compression” available Jul 5, 2022 (hereinafter “Zhirui”) in further view of Benjamin et al; “Automated Quantum Hardware Selection for Quantum Workflows” available Apr 20, 2021 (hereinafter “Benjamin”).
Regarding claim 8, Jay does not teach The system of claim 5, wherein the at least one of the computer-executable components further: prioritizes analysis of the set of quantum computing systems according to at least one of system availability, qubit count, noise levels, or coherence times.
However, Benjamin does teach The system of claim 5, wherein the at least one of the computer-executable components further: prioritizes analysis of the set of quantum computing systems according to at least one of system availability, qubit count, noise levels, or coherence times. (Benjamin Section 4.2 Paragraph 5; "Subsequently, the NISQ Analyzer uses up-to-date provenance data about the quantum hardware, such as the number of qubits or their decoherence Times" Examiner notes that the NISQ Analyzer is prioritize analysis of the set of quantum computing systems according to qubit count or decoherence times)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Benjamin. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. Benjamin teaches quantum hardware selection for quantum workflows by analyzing the quantum circuits. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Benjamin to ensure quantum algorithm can be performed on quantum hardware “to determine which quantum hardware can successfully execute the given quantum circuit [LU,3J].” (Benjamin Page 9 Paragraph 1).
Regarding claim 16, Jay does not teach The computer-implemented method of claim 13, wherein the device prioritizes analysis of the set of quantum computing systems according to at least one of system availability, qubit count, noise levels, or coherence times.
However, Benjamin does teach The computer-implemented method of claim 13, wherein the device prioritizes analysis of the set of quantum computing systems according to at least one of system availability, qubit count, noise levels, or coherence times. (Benjamin Section 4.2 Paragraph 5; "Subsequently, the NISQ Analyzer uses up-to-date provenance data about the quantum hardware, such as the number of qubits or their decoherence Times" Examiner notes that the NISQ Analyzer is prioritize analysis of the set of quantum computing systems according to qubit count or decoherence times)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Benjamin. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. Benjamin teaches quantum hardware selection for quantum workflows by analyzing the quantum circuits. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Benjamin to ensure quantum algorithm can be performed on quantum hardware “to determine which quantum hardware can successfully execute the given quantum circuit [LU,3J].” (Benjamin Page 9 Paragraph 1).
Claim(s) 19 is rejected under 35 U.S.C. 103 as being unpatentable over Jay et al; US 10831455 B2 published Nov 10, 2020 (hereinafter “Jay”) in view of Weicheng Kong; US 20240005190 A1 filed on Sep 29, 2021 (hereinafter “Kong”) in further view of Nathon Wiebe et al; US 20200279185 A1filed on Feb 28, 2019 (hereinafter “Wiebe”) in further view of Jared Hertzberg et al; US 20200401925 A1 published on Dec 24, 2020 (hereinafter “Hertzberg”) in further view of in further view of Glushakov et al; US 20210349914 A1 filed on Jul 20, 2021 (hereinafter “Glushakov”) in further view of Aaron Baughman; “Study of Feature Importance for Quantum Machine Learning Models” available Jun 9, 2022 (hereinafter “Aaron”) in further view of Zhirui et al; “Quantum Neural Network Compression” available Jul 5, 2022 (hereinafter “Zhirui”) in further view of UZDIN RAAM et al; WO 2023181019 A1 filed (hereinafter “Uzdin”).
Regarding Claim 19, Jay does not teach The computer program product of claim 17, wherein the at least one processor is configured to adjust the trained quantum machine learning model by iteratively removing, in random order, at least one logical qubit or at least one quantum gate from the trained quantum machine learning model.
However, Uzdin does teach The computer program product of claim 17, wherein the at least one processor is configured to adjust the trained quantum machine learning model by iteratively removing, in random order, at least one logical qubit or at least one quantum gate from the trained quantum machine learning model. (Uzdin Page 46 line 20; “Each RC realization is obtained by randomly replacing and/or adding some gates that leave the noise-free target circuit invariant.” Examiner notes that this citation is supported by at least Figure 2(a) and Page 5 Paragraph 2 of Uzdin’s provisional application; replacing a gate requires it to be removed first; each RC realization that removes a gate is an iteration)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, Benjamin, and Uzdin. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. Uzdin teaches a method for quantum error mitigation of noise in a quantum system. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, Benjamin, and Uzdin to select the appropriate quantum computing system for a quantum machine learning model based on the qubit coupling maps and adjust the quantum machine learning model to perform on other quantum computing systems “Randomized compiling (RC) is a standard method that allows to transform these errors into stochastic noise, which is more amenable to the application of QEM.” (Uzdin Page 46 line 20)
Claim(s) 20 is rejected under 35 U.S.C. 103 as being unpatentable over Jay et al; US 10831455 B2 published Nov 10, 2020 (hereinafter “Jay”) in view of Weicheng Kong; US 20240005190 A1 filed on Sep 29, 2021 (hereinafter “Kong”) in further view of Nathon Wiebe et al; US 20200279185 A1filed on Feb 28, 2019 (hereinafter “Wiebe”) in further view of Jared Hertzberg et al; US 20200401925 A1 published on Dec 24, 2020 (hereinafter “Hertzberg”) in further view of in further view of Glushakov et al; US 20210349914 A1 filed on Jul 20, 2021 (hereinafter “Glushakov”) in further view of Aaron Baughman; “Study of Feature Importance for Quantum Machine Learning Models” available Jun 9, 2022 (hereinafter “Aaron”) in further view of Zhirui et al; “Quantum Neural Network Compression” available Jul 5, 2022 (hereinafter “Zhirui”) in further view of Gusat; Mircea R. et al; US 20230325269 A1 (hereinafter “Gusat”).
Regarding Claim 20, Jay does not teach The computer program product of claim 17, wherein the set of variable importance scores are based on at least one of univariate Fisher scores of the trained quantum machine learning model or local scope Shapley values of the trained quantum machine learning model.
However, Conort does teach The computer program product of claim 17, wherein the set of variable importance scores are based on at least one of univariate Fisher scores of the trained [quantum] machine learning model or local scope Shapley values of the trained [quantum] machine learning model. (Conort Paragraph 0133; “The Shapley value may indicate an importance (e.g., overall contribution, level of contribution, or amount of contribution) of each feature within the overall prediction of a machine learning model” Examiner notes that set of variable importance scores are based on local scope Shapley values of the trained machine learning model (Shapley value of each feature of a machine learning model))
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Conort. Jay teaches selected a quantum device to execute the set of quantum assembly language. Kong teaches quantum topology graph optimization. Wiebe teaches methods to train a quantum Boltzmann machine (QBM) having one or more visible nodes and one or more hidden nodes. Hertzberg teaches methods that can facilitate quantum circuit topology selection based on frequency collisions between qubits. Glushakov teaches comparing metric graphs using Gromov-Hausdorff distance. Aaron teaches the application of feature importance for quantum machine learning models. Zhirui teaches quantum neural network pruning. Conort teaches automated feature engineering for machine learning models. One of ordinary skill would have motivation to combine Jay, Kong, Wiebe, Hertzberg, Glushakov, Aaron, Zhirui, and Conort to select the appropriate quantum computing system for a quantum machine learning model based on shapely values to incorporate its desirable properties “SHAP has a number of desirable properties that its precursors lacked: (i) local accuracy of the explanation model prediction; (ii) missingness—features missing in the original input must have no impact; and (iii) consistency when revising the original model.” (Gusat Paragraph 0062)
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 DANIEL DUC TRAN whose telephone number is (571)272-6870. The examiner can normally be reached Mon-Fri 8:00-5:00 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 at (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.
/D.D.T./Examiner, Art Unit 2147
/ERIC NILSSON/Primary Examiner, Art Unit 2151
Prosecution Insights