DETAILED ACTION
Notice of Pre-AIA or AIA Status
1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
2. Claims 1-20 are presented for examination.
Claim Rejections - 35 USC § 101
3. 35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
3.1 Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 2A- Prong One
The claim(s) recite(s) a method, system, and computer program product, comprising: The step of: “generates a different dimensional parameterization based on the input parameterization, the different dimensional parameterization for inputting to a physical model that approximates the physical system; running the physical model using the different dimensional parameterization”, “generates an output solution based on the different dimensional parameterization input to the physical model”, “based on the output solution and the target property, training the neural network to generate the different dimensional parameterization”, under the broadest reasonable interpretation fall under a mental process or otherwise could fall under a mathematical concept / mathematical relationship. Therefore, the claims are directed to an abstract idea, by use of generic computer components and thus are clearly directed to an abstract idea, as constructed.
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional limitation such as: “at least one processor”, “a memory”, a computer-readable … medium”, “program instructions”.. readable by “a device”, either alone or in combination, all serve to gather and process data and do not add anything more significantly to the judicial exception, but are mere instructions to apply the exception using a generic computer component that are well known, routine, and conventional activities (see specification at para [0017-0020], which can be of any type , including general-purpose computer (para [0022] PROCESSOR SET 110 includes one, or more, computer processors of any type now known) previously known in the industries. Merely adding a programmable computer to perform generic computer functions does not automatically overcome an eligibility rejection. Alice, 573 U.S. at 223-24. Furthermore, the use of a general-purpose computer to apply an otherwise ineligible algorithm does not qualify as a particular machine. See Ultramerciallnc. v. Hulu, LLC, 772F.3d 709, 716-17 (Fed. Cir. 20l4); In re TLI Commc 'ns LLC v. AV Automotive, LLC, 823 F.3d 607, 613 (Fed. Cir. 2016) (mere recitation of concrete or tangible components is not an inventive concept); Eon Corp. IP Holdings LLC v. AT&T Mobility LLC, 785; the step of: “receiving a parameterization of a physical system, the physical system including real physical components, the parameterization having corresponding target property in the physical system”; “inputting the parameterization into a neural network”, under the broadest reasonable interpretation, reasonable fall under data gathering and processing activities that are pre-solution activities” are also well-known, routine and conventional activities and are not sufficient to amount to significantly more than the judicial exception (See further MPEP 2106.05(d)(i-iv)-f); thus are not patent eligible under 35 USC 101.
Step 2B
The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as previously discussed above with reference to the integration of abstract idea into a practical application, the additional elements of: “at least one processor”, “a memory”, a computer-readable … medium”, “program instructions”... readable by “a device”, either alone or in combination, all serve to gather and process data and do not add anything more significantly to the judicial exception, but are mere instructions to apply the exception using a generic computer component that are well known, routine, and conventional activities (see specification at para [0017-0020], which can be of any type , including general-purpose computer (para [0020] PROCESSOR SET 110 includes one, or more, computer processors of any type now known) previously known in the industries. Merely adding a programmable computer to perform generic computer functions does not automatically overcome an eligibility rejection. Alice, 573 U.S. at 223-24. Furthermore, the use of a general-purpose computer to apply an otherwise ineligible algorithm does not qualify as a particular machine. See Ultramerciallnc. v. Hulu, LLC, 772F.3d 709, 716-17 (Fed. Cir. 20l4); In re TLI Commc 'ns LLC v. AV Automotive, LLC, 823 F.3d 607, 613 (Fed. Cir. 2016) (mere recitation of concrete or tangible components is not an inventive concept); Eon Corp. IP Holdings LLC v. AT&T Mobility LLC, 785; the step of: “receiving a parameterization of a physical system, the physical system including real physical components, the parameterization having corresponding target property in the physical system”; “inputting the parameterization into a neural network”, under the broadest reasonable interpretation, reasonable fall under data gathering and processing activities that are pre-solution activities” are also well-known, routine and conventional activities and are not sufficient to amount to significantly more than the judicial exception (See further MPEP 2106.05(d)(i-iv)-f); thus are not patent eligible under 35 USC 101. Therefore, using computer components amount to no more than mere instructions to perform the abstract, and thus are not sufficient to amount to significantly more than the recited abstract, as constructed.
3.2 Dependent claims 2-12, 14-17, 19-20 merely include limitations pertaining to further mathematical computations (claims 2, 14, 19), “wherein the training of the neural network includes iterating at least: updating parameters of the neural network; running the neural network with the updated parameters for the neural network to generate the different dimensional parameterization; and the running of the physical model using the different dimensional parameterization; wherein the iterating is performed until a threshold convergence in an error between the output solution and the target property is reached” (mathematical concept). (claims 3 and 15); “wherein the target property is generated by running another physical model, which has higher fidelity than the physical model” (mental process); (claims 4 and 16); “wherein the physical model simulates in coarser resolution said another physical model” (mental process); (claim 5); “wherein the physical model omits a portion of physical processes in said another physical model” (mental process); (claim 6) “wherein the physical model collapses at least one dimension used in said another physical model” (mental process); (claim 7); “wherein the physical model is a discretization of said another physical model” (mental process); (claim 8) “wherein the target property is generated from experimental data” (mental process); (claim 9) “wherein the different dimensional parameterization has coarser resolution than the received parameterization of the physical system” (mental process); (claims 10, 17, and 20) “obtaining a down-sampled version of the received parameterization, and wherein weighted combination of the down-sampled version of the received parameterization and the different dimensional parameterization output by the neural network is input to the physical model” (data gathering and processing or otherwise could fall under mental process); (claim 11) “wherein weights used in the weighted combination are learned” (mental process); “wherein the neural network imposes symmetry constraints on the generated different dimensional parameterization” (mental process), all of which further amount to further mathematical concept and/or mental process similar to that already recited by the independent claims and already addressed above and thus are further not patent eligible under 35 USC 101.
Claim Rejections - 35 USC § 103
4. 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.
5. Claim(s) 1-2, 8, 12, 14, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Thierauf et al. (USPG_PUB No. 2021/0116899), in view of Lam et al. (USPG_PUB No. 2014/0365180).
5.1 In considering claims 1, 13, and 18, Thierauf et al. teaches a method comprising:
receiving a parameterization of a physical system, the physical system including real physical components, the parameterization having corresponding target property in the physical system (para [0012], the method can provide, e.g., a basic parameterization for simple components and/or a specifically adapted parameterization and/or optimization of the basic parameterization for selected components (e.g., a control node, like a SPS/PLC). [0025], The parameterization unit comprises: a parameter interface for reading a basic parameterization); inputting the parameterization into a neural network (see para [0025], an interface to a machine learning module, via which a result message can be read e.g. as input in to adapt the basic parameterization with a target parameterization. A processor for controlling and operating the component with the basic parameterization and/or target parameterization. [0048], All or selected components K can include a parameterization unit P, as shown in FIG. 1 as an example for the components K1 and K2. [0053], The basic parameterization is usually received and read by a server S. During the test run, measured values are measured and read in or recorded and transmitted to the network ANN in order to calculate the target parameterization), wherein the neural network generates a different dimensional parameterization based on the input parameterization, the different dimensional parameterization for inputting to a physical model that approximates the physical system (see para [0008], a pre-trained neural network, wherein the pre-trained neural network (hereinafter also referred to as ANN for ‘artificial neural network’) is pre-trained in order to calculate a target parameterization for the respective component for a measured value data record. [0031] The machine learning module is an electronic unit that can be trained in software and/or hardware. The machine learning module can include a pattern recognition algorithm in one version, which is applied to the acquired measurement data. The machine learning module can comprise a pre-trained or trained neural network and, if necessary, other machine learning methods. The machine learning module can be trained on a server and can be connected to a database. [0032] A trained neural network, also known as artificial neural network (ANN), is a computer-implemented method for calculating an optimized parameter data set (target parameterization). The ANN is based on training data obtained from laboratory and/or simulation data. The training data comprise input data and output data. The input data comprise the measured value data acquired from one component each. The output data comprise “optimal” parameterization.); and based on the output solution and the target property, training the neural network to generate the different dimensional parameterization (see para [0008], a pre-trained neural network, wherein the pre-trained neural network (hereinafter also referred to as ANN for ‘artificial neural network’) is pre-trained in order to calculate a target parameterization for the respective component for a measured value data record. [0011], the machine learning module is continuously re-trained with the continuously recorded measured value data set, the basic parameterization and the calculated target parameterization. [0032], The network is trained on the basis of this training data in such a way that it calculates the target parameterization for any measured value data of a component. [0047], The machine learning module ML can be trained to generate a model for the parameterization of components K. The generated model can be stored in a database DB.). However, he does not expressly show running the physical model using the different dimensional parameterization, wherein the physical model generates an output solution based on the different dimensional parameterization input to the physical model.
Lam et al. teaches the step of running the physical model using the different dimensional parameterization, wherein the physical model generates an output solution based on the different dimensional parameterization input to the physical model (see para [0026], In one aspect, no physical simulation need be performed when evaluating the response surface; Physical simulation is only performed during the construction of the response surface. For instance, the response surface takes a number of sets of inputs and corresponding outputs obtained by running the physical models. Then a model similar to regression models may be built to capture the relationship between the inputs and outputs.). Thierauf et al. AND Lam et al. are analogous art because they are from the same field of endeavor and that the model analyzes by Lam et al. is similar to that of Thierauf et al. Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Lam et al. with that Thierauf et al. because Lam et al. teaches maximizing improvement at a cost saving (see [0045]).
5.2 As per claims 2, 14, and 19, the combined teachings of Thierauf et al. and Lam et al. teaches that wherein the training of the neural network includes iterating at least: updating parameters of the neural network and running the neural network with the updated parameters for the neural network to generate the different dimensional parameterization (see Thierauf et al. para [0051], A machine learning model pre-trained for the respective component K thus analyzes the acquired sensor data or measured values and tries to recognize certain patterns that indicate optimization potential. Based on the anomalies found, recommendations are displayed in the form of a result message to the user (operator of the AA machine with the respective component K). On the basis of the results found, recommendations are displayed to the user in order to optimize parameters. If the target parameterization differs from the basic parameterization, further steps can be initiated by accessing a set of rules (which can be stored, e.g., in a rule base). For example, if a deviation is considered relevant, a result message can be created that includes a set of commands that initiate or instruct an adjustment of the basic parameterization on component K. Afterwards the method can end or be repeated after a predetermined time unit. For example, the process can be repeated in order to check the executed parameter changes. Further Lam et al. para [0041], The surrogate model is refit to the augmented data, and the steps of optimization, model simulation and surrogate model rebuilding are iterated until one of several stopping criteria is met. A stopping criterion may be that the expected improvement is smaller than a threshold value. For example, if the expected improvement of a new variable combination is (a) less than a small fraction t.sub.a (e.g., 1%) of the current minimum EUI or (b) smaller than a pre-defined meaningful threshold t.sub.b (e.g., 0.05), the search terminates. For practical reasons and limitations on total computation time of the entire search, the search may also be stopped if it has not resulted in any actual improvement of EUI in a given number of simulations, or has exceeded an acceptable number of iterations.); and the running of the physical model using the different dimensional parameterization, wherein the iterating is performed until a threshold convergence in an error between the output solution and the target property is reached (see Lam et al. para [0026], For instance, the response surface takes a number of sets of inputs and corresponding outputs obtained by running the physical models. Then a model similar to regression models may be built to capture the relationship between the inputs and outputs [0045], the simulation model 106 is run with the additional design points 116. The additional design points 116 are those that maximize the expected improvement. The energy use estimation at the new design point 118 output from the simulation 106 is used with the previous output 108 in modeling parameters in the Gaussian process, and the surrogate model 102 is updated or rebuilt with this new input-output relationship of the Gaussian process. On the other hand, if the response surface of expected improvement function for cost saving over the considered time period 114 converges, the last set of input combination is output 120 or recommended as an optimum combination of building components. If it is determined that the improvement is converging, at 312, the design points used in the latest simulation are output as recommended combination of building components. If it is determined that the improvement is not converging, the logic of the method returns to 304, where additional simulation points are designed, and the processing at 306, 308, 310 are repeated. The energy performance simulation results at additional design points may then be used to update the surrogate model.). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Lam et al. with that Thierauf et al. because Lam et al. teaches maximizing improvement at a cost saving (see [0045]).
5.3 Regarding claim 8, the combined teachings of Thierauf et al. and Lam et al. teaches that wherein the target property is generated from experimental data (see Lam et al. para 0045] FIG. 1 illustrates an overview of a statistics optimization methodology of the present disclosure in one embodiment. As described in detail above, a surrogate model 102 may be built based on initial design and also using a sequential design technique. The surrogate model 102 may be initially built using a space-filling design 104 of experiments in the input space for initial planning of an energy simulation. The initial design may comprise space-filling design of experiments in the input space 104, which is used to run a building simulation model 106, which in turn produces the energy use estimation at those design points 108. [0047], . At 202, initial design of experiments at various combinations of building product properties may be created by space filling design. [0032], The training data comprise input data and output data. The input data comprise the measured value data acquired from one component each. The output data comprise “optimal” parameterization. The network is trained on the basis of this training data in such a way that it calculates the target parameterization for any measured value data of a component.). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Lam et al. with that Thierauf et al. because Lam et al. teaches maximizing improvement at a cost saving (see para [0045]).
5.12 With regards to claim 12, the combined teachings of Thierauf et al. and Lam et al. teaches that wherein the neural network imposes symmetry constraints on the generated different dimensional parameterization (see Lam et al. constraints optimization at para [0037], The constrained optimization for S.sub.i may be performed, e.g., using a computer-implemented statistical tool (e.g., R Core Team), for instance, constrOptim function in R.). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Lam et al. with that Thierauf et al. because Lam et al. teaches maximizing improvement at a cost saving (see [0045]).
6. Claim(s) 3-7, 9-11, 13, 15-18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Thierauf et al. (USPG_PUB No. 2021/0116899), in view of Lam et al. (USPG_PUB No. 2014/0365180), further in view of Abrahamson et al. (USPG_PUB No. 20170011143).
6.1 With regards to claims 3 and 15, the combined teachings of Abrahamson et al. and Lam et al. teaches most of the instant invention, including wherein the target property is generated by running another physical model (using an optimal envelope design with a high dimensional variable space, para [0023], [0026], …, the response surface takes a number of sets of inputs and corresponding outputs obtained by running the physical models. Then a model similar to regression models may be built to capture the relationship between the inputs and outputs); however, he does not specifically state the use of a higher fidelity model than the physical model. Abrahamson et al. teaches a method for solving multi-fidelity optimization problems (see title), including simulation model having higher fidelity than the physical model (see abstract, para [0017] At block 130, the reduced number of trial points is then evaluated by the objective function and any constraints with a high fidelity simulation. The high fidelity simulation is run only once per search step. [0029], Each point in the preliminary set V.sub.o is evaluated by the objective function and any constraints, which call the high fidelity simulation, yielding the preliminary set V.sub.0 of evaluated trial points.). Thierauf et al., Lam et al., and Abrahamson et al. are analogous art because they are from the same field of endeavor and that the model analyzes by Abrahamson et al. is similar to that of Thierauf et al. and Lam et al. Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Abrahamson et al. with that of Thierauf et al. and Lam et al. because Abrahamson et al. provides a reduce number of trials in the solution (see par [0051]).
6.2 As per claims 4 and 16, the combined teachings of Thierauf et al., Lam et al., and Abrahamson et al. teaches that wherein the physical model simulates in coarser resolution said another physical model. Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Abrahamson et al. with that of Thierauf et al. and Lam et al. because Abrahamson et al. provides a reduce number of trials in the solution (see par [0051]).
6.3 Regarding claim 5, the combined teachings of Thierauf et al., Lam et al., and Abrahamson et al. teaches that wherein the physical model omits a portion of physical processes in said another physical model (see Lam et al. para [0034), For a design X of n runs, define X.sub.i to be a design of n-1 points (excluding the i th point), and define X.sub.i(x) as design X.sub.i augmented by a new input x. Therefore the minimum distance of all the design points .phi.sub..lamda .(X.sub.i(x)) (Equation 3) is a function of x only, as the remaining n-1 points are fixed. Now n reduced S.sub.i problems (the problem of finding the x.sub.i that maximize the minimum distance of all the points in the initial design space). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Abrahamson et al. with that of Thierauf et al. and Lam et al. because Abrahamson et al. provides a reduce number of trials in the solution (see par [0051]).
6.4 With regards to claim 6, the combined teachings of Thierauf et al., Lam et al., and Abrahamson et al. teaches that wherein the physical model collapses at least one dimension used in said another physical model (see Abrahamson et al. para [0028] The low fidelity simulations are usually simplified physics models of the high-fidelity simulations. As one simple example, a high fidelity simulation describes three-dimensional flow of air over an airfoil, a lower fidelity simulation describes two-dimensional flow, and the lowest fidelity simulation describes one-dimensional flow.). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Abrahamson et al. with that of Thierauf et al. and Lam et al. because Abrahamson et al. provides a reduce number of trials in the solution (see par [0051]).
6.5 As per claim 7, the combined teachings of Thierauf et al., Lam et al., and Abrahamson et al. teaches that wherein the physical model is a discretization of said another physical model (see Abrahamson et al. para [0028] The low fidelity simulations are usually simplified physics models of the high-fidelity simulations so as to discretize the model. As one simple example, a high fidelity simulation describes three-dimensional flow of air over an airfoil, a lower fidelity simulation describes two-dimensional flow, and the lowest fidelity simulation describes one-dimensional flow.). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Abrahamson et al. with that of Thierauf et al. and Lam et al. because Abrahamson et al. provides a reduce number of trials in the solution (see par [0051]).
6.6 With regards to claim 9, the combined teachings of Thierauf et al., Lam et al., and Abrahamson et al. teaches that wherein the different dimensional parameterization has coarser resolution than the received parameterization of the physical system (see Abrahamson et al. para [0028] The low fidelity simulations are usually simplified physics models of the high-fidelity simulations so as to discretize the model. As one simple example, a high fidelity simulation describes three-dimensional flow of air over an airfoil so to coarsen the model, a lower fidelity simulation describes two-dimensional flow, and the lowest fidelity simulation describes one-dimensional flow.). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Abrahamson et al. with that of Thierauf et al. and Lam et al. because Abrahamson et al. provides a reduce number of trials in the solution (see par [0051]).
6.7 As per claims 10, 17, and 20, the combined teachings of Thierauf et al., Lam et al., and Abrahamson et al. teaches the step of obtaining a down-sampled version of the received parameterization, and wherein weighted combination of the down-sampled version of the received parameterization and the different dimensional parameterization output by the neural network is input to the physical model (see Abrahamson et al. para [0041] At block 250, an update is performed. For instance, mesh size and barrier parameter are updated. The surrogate solution is mapped to its nearest mesh point. If the search step is unsuccessful, the poll step may reduce the mesh size su as to down-sample the data. If the search step is successful, the mesh size may instead be increased.; Lam et al. para [0030-0031] With our example problem setting of 15 parameters, we will use 100 set of inputs as initial design. The following will describe the method to obtain the initial design of 100 points as an example. Briefly, design points refer to sample input points to the simulation model, e.g., various combinations of building product components. A space-filling design e.g. weight function is used in one embodiment of the present disclosure for initial planning of an energy simulation. [0039], Red dots represent the observed responses that have been obtained by evaluating f at selected sample points. Based on the fitted response surface (dotted line) only, one might expect the minimum to be at x=9.4. further see Thierauf et al. para [0032] A trained neural network, also known as artificial neural network (ANN), is a computer-implemented method for calculating an optimized parameter data set (target parameterization). The ANN is based on training data obtained from laboratory and/or simulation data. The training data comprise input data and output data. The input data comprise the measured value data acquired from one component each. The output data comprise “optimal” parameterization. The network is trained on the basis of this training data in such a way that it calculates the target parameterization for any measured value data of a component.). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Abrahamson et al. with that of Thierauf et al. and Lam et al. because Abrahamson et al. provides a reduce number of trials in the solution (see par [0051]).
6.8 Regarding claim 11, the combined teachings of Thierauf et al. and Lam et al. teaches that wherein weights used in the weighted combination are learned (see Abrahamson et al. para [0049], It is desired to minimize the weight, subject to safety margin constraints on stress, displacement, buckling, etc. Constraints also include minimum thickness (lower bound) constraints on each component. The weight may be inexpensive to compute, since weight is the product of volume and density, the density of each material is known; Thierauf et al. para [0011], the machine learning module is continuously re-trained with the continuously recorded measured value data set, the basic parameterization and the calculated target parameterization. The machine learning module thus stores a model which is self-learning and is fed with continuously newly measured data and continuously “continues learning”. The model can be used advantageously for other systems and/or for creating a basic parameterization.). Therefore, it would have been obvious to a person of skilled in the art at the time of filing of the applicant’s invention to combine the method of Abrahamson et al. with that of Thierauf et al. and Lam et al. because Abrahamson et al. provides a reduce number of trials in the solution (see par [0051]).
Conclusion
7. Claims 1-20 are rejected and this action is non-final. Any inquiry concerning this communication or earlier communications from the examiner should be directed to ANDRE PIERRE-LOUIS whose telephone number is (571)272-8636. The examiner can normally be reached M-F 9:00 AM-5:00 PM.
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, EMERSON C PUENTE can be reached at 571-272-3652. 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.
/ANDRE PIERRE LOUIS/Primary Patent Examiner, Art Unit 2187 April 4, 2026
Prosecution Insights