Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 29 July 2026 has been entered.
Claim Rejections - 35 USC § 101
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.
Claims 1-8 and 10-21 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a mental process without significantly more.
Independent claim 1 recites:
Claim 1. A computer-implemented method for detecting performance regressions in a database system comprising:
receiving time series data representing performance test results for a database system;
determining a set of candidate change points comprising a plurality of pre-change points in the time series data by applying a Bayesian model to the time series data;
determining one or more change points from among the set of candidate change points by applying a Pruned Exact Linear Time (PELT) model to the time series data and the plurality of pre-change points, wherein applying the PELT model to the time series data and the plurality of pre-change points comprises:
determining respective costs of the plurality of pre-change points using a penalized cost function; and
minimizing the penalized cost function over the time series data; and outputting the one or more change point; and
outputting the one or more change points to a benchmark monitor, wherein the benchmark monitor is configured to receive the one or more change points as inputs and capture one or more performance metrics for performance regression tests;
receiving user input generated via a manual verification process for at least one of the one or more change points; and
adjusting, based on the user input, at least one of a prior distribution of the Bayesian model or a penalty value parameter of the PELT model.”
Claims 12 and 17 contain similar subject matter.
This is a mental process because the claimed method analyzes change-point data to “determining a set of candidate change points comprising a plurality of pre-change points in the time series data by applying a Bayesian model to the time series data,” “determining one or more change points from among the set of candidate change points by applying a Pruned Exact Linear Time (PELT) model to the time series data and the plurality of pre-change points” and “adjusting, based on the user input, at least one of a prior distribution of the Bayesian model or a penalty value parameter of the PELT model.” The method then describes additional mental operations using the PELT model, comprising: “determining respective costs of the plurality of pre-change points using a penalized cost function;” and “minimizing the penalized cost function over the time series data; and outputting the one or more change points.” Claim 12 adds that the PELT model “incorporates a Radial Basis Function (RBF) model.” Claim 17 adds “adjusting one or more parameters of the Bayesian model and/or the PELT model based on the user input.” Each of these are similarly mental processes or data analysis definitions.
These are each mental process steps that are data analysis and data definitions. A human being equipped with pen and paper or a generic machine would be capable of following these steps to analyze time series data.
The claims contain additional elements beyond the mental process in the form of “receiving time series data…,” “stor[ing a] Bayesian model and [a Pruned Extract linear Time model that incorporates a Radial Basis Function (RBF) model,” “at least one hardware processor” and “at least one memory coupled to the at least one hardware processor,” (claim 12); “one or more non-transitory computer readable media,” (claims 12 and 17); “outputting the one or more change points,” and “receiving user input regarding the one or more change points,” “wherein the benchmark monitor is configured to receive the one or more change puts as input and capture one or more performance metrics for performance regression tests,” and “receiving user input generated via a manual verification process for at least one of the one or more change points.”
This judicial exception is not integrated into a practical application because the claimed additional elements do not appear to improve the processing of a computer, require the use of a specific machine, effect a transformation or reduction of a particular article to a different state or thing, or provide a technological solution to a technological problem.
The “at least one hardware processor,” “at least one memory coupled to the at least one hardware processor,” and “one or more non-transitory computer readable media” appear to be generic computing hardware elements. The recitation of generic hardware is little more than using a computer to perform an abstract idea, see MPEP 2106.05(f)(2). The “Receiving…” steps appear to be data gathering steps and are thus mere pre-solution insignificant activity (see MPEP 2106.05(g). Storing the analysis models in a memory is similarly insignificant extra-solution activity and is well-known (see MPEP 2106.05(d)(II) and MPEP 2106.05(g)). Outputting of a data analysis by is insignificant post-solution activity (see MPEP 2106.05(g)(3)).
It is noted that none of the additional elements appear to improve the processing of a computer, require the use of a specific machine, effect a transformation or reduction of a particular article to a different state or thing, or provide a technological solution to a technological problem. As such, none of the additional elements appear to integrate the judicial exception into a practical application.
None of the additional elements are sufficient to amount to significantly more than the judicial exception, in part or in whole.
The recitation of generic hardware of “at least one hardware processor,” “at least one memory coupled to the at least one hardware processor,” and “one or more non-transitory computer readable media” is little more than using a computer to perform an abstract idea, see MPEP 2106.05(f)(2). The additional element of the “receiving…” steps are merely extra-solution activity data gathering and is well understood, routine, and conventional (see MPEP 2106.05(g)). Storing the data models in a memory is nothing more than storing data in a memory, which is recognized as well-understood, routine, and conventional (see MPEP 2106.05(d)(II)). Displaying an output of a data analysis is insignificant extra-solution activity and is well known (see MPEP 2106.05(g)(3)).
None of the additional elements, in part or in whole, appear to improve the processing of a computer, require the use of a particular machine, effect a transformation or reduction of a particular article to a different state or thing, or add a specific limitation other than what is well understood, routine, or conventional. As such, none of the additional elements appears to be, in part or as a whole, significantly more than the judicial exception.
Dependent claims 2-11, 13-16, and 18-20 are merely directed towards additional limitations that further define data types and data models or further describe analyses that will occur. It is noted that the claimed data definitions and data analysis steps do not appear to include additional elements that incorporate the claimed subject matter into a practical application. The dependent claims also do not include additional elements that, in part or in whole, appear to be significantly more than the abstract idea.
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.
Claims 1-2, 5-8, 10-11 and 12-20 are rejected under 35 U.S.C. 103 as being unpatentable over Liu et al. (US Patent 12,306,811) in view of Kaçar et al. (“Automatic Segmentation of Time Series Data with PELT Algorithm for Predictive Maintenance in the Flat Steel Industry”), and further in view of Hamm (US Pre-Grant Publication 2025/0053823).
As to claim 1, Liu et al. teaches a computer-implemented method for detecting performance regressions in a database system, comprising:
receiving time series data representing performance test results for a database system (see Liu 2:63-3:17. Liu is directed towards changepoint detection in a time series database. It is noted that the application may include various “test results,” such as those listed in Liu 3:6-19);
determining a set of candidate change points comprising a plurality of pre-change points in the time series data by applying a Bayesian model to the time series data (see Liu 3:28-32. The system may identify probable, or candidate, changepoints using a Bayesian model);
determining one or more change points from among the set of candidate change points by applying a Pruned Exact Linear Time (PELT) model to the time series data and the plurality of pre-change points (see Liu 3:36-41. The system may use a PELT model to identify changepoints. It is noted that Liu teaches in 15:8-16 that Liu may use “one or more models” for changepoint detection. Also see Liu 19:28-42 and 21:33-45, which explicitly discuss how output from a first model may be used as input to a second model to confirm statistical analysis. It is noted that 21:33-45 shows that a first changepoint has been detected and a second model may be used to analyze the first changepoint data. This is also shown in claims 2 and 10 of Liu),
…
outputting the one or more change points (see Liu 19:28-42).
Liu does not explicitly teach:
wherein applying the PELT model to the time series data and the plurality of pre-change points comprises:
determining respective costs of the plurality of pre-change points using a penalized cost function; and
minimizing the penalized cost function over the time series data; and
outputting the one or more change points to a benchmark monitor, wherein the benchmark monitor is configured to receive the one or more change points as inputs and capture one or more performance metrics for performance regression tests;
receiving user input generated via a manual verification process for at least one of the one or more change points; and
adjusting, based on the user input, at least one of a prior distribution of the Bayesian model or a penalty value parameter of the PELT model.
Kaçar teaches:
wherein applying the PELT model to the time series data and the plurality of pre-change points comprises:
determining respective costs of the plurality of pre-change points using a penalized cost function (see Kaçar pages 903-904, Section III and III A. PELT operates by inputting a data model and using a penalized cost function); and
minimizing the penalized cost function over the time series data (see Kaçar pages 903-904, Section III and III A. PELT functions operate to minimize a cost function).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Liu by the teachings of Kaçar because both references are directed towards statistical analysis of data for the identification of change points using PELT and because Kaçar provides to Liu additional considerations to incorporate into the PELT algorithm that will further minimize costs.
Hamm teaches outputting the one or more change points to a benchmark monitor, wherein the benchmark monitor is configured to receive the one or more change points as inputs and capture one or more performance metrics for performance regression tests (see Hamm paragraphs [0022], [0030], and [0051]. Hamm shows that change points may be output a monitor for user feedback and testing a variety of performance metrics. It is noted that “for performance regression tests” is an intended use and receives no patentable weight);
receiving user input generated via a manual verification process for at least one of the one or more change points (see Hamm paragraphs [0022], [0030], and [0051]. Hamm shows wherein users may input feedback regarding the identification of changepoints. This feedback may then be used to retrain the model); and
adjusting, based on the user input, at least one of a prior distribution of the Bayesian model or a penalty value parameter of the PELT model (see Hamm paragraphs [0018], [0022], [0030], and [0051]. User input may be used as a performance metric to update a bias metric to retrain a Bayesian model).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Liu by the teachings of Hamm because both references are directed towards statistical analysis of data for the identification of change points and because Hamm provides to Liu additional input that will allow a model to be improved over time, increasing the utility to a user and the accuracy of a model.
As to claim 2, Liu teaches the method of claim 1, wherein applying the Bayesian model to the time series data comprises determining respective Bayes factors for positions in the time series data (see Liu 3:28-32).
As to claim 5, Liu as modified by Kaçar teaches the method of claim 1, wherein:
the PELT model incorporates a Radial Basis Function (RBF) model (see Kaçar page 904, Section III A. The cost function may use a Radial Basis Function);
applying the PELT model to the time series data and the plurality of pre-change points further comprises defining expected variations between data points in different segments of the time series data using the RBF model (see Kaçar page 904, Section III A); and
the determination of the respective costs is based at least in part on the expected variations (see Kaçar page 904, Section III A).
As to claim 6, Liu as modified teaches the method of claim 1, further comprising receiving a distribution type of the time series data, wherein the determination of the plurality of pre-change points is based at least in part on the distribution type (see Liu 3:6-19).
As to claim 7, Liu as modified by Kaçar teaches the method of claim 1, further comprising receiving a penalty value, wherein the determination of the one or more change points from among the set of candidate change points is based at least in part on the penalty value (see Kaçar page 904, Section III A).
As to claim 8, Liu as modified teaches the method of claim 1, wherein the time series data comprises pre-processed time series data which has undergone pre-processing to reduce noise in the time series data and remove outliers from the time series data (see Liu 14:60-15:7).
As to claim 10, Liu as modified by Hamm teaches the method of claim 1, wherein the user input comprises at least one of the following for a given change point of the one or more change points:
an indication to designate the given change point as a confirmed change point (see Hamm paragraphs [0022], [0030], and [0051]);
an indication to remove the given change point;
an indication to designate the given change point as a pending change point for further review (see Hamm paragraphs [0022], [0030], and [0051]); or
an indication to adjust a position of the given change point and designate the given change point as a modified change point (see Hamm paragraphs [0022], [0030], and [0051]).
As to claim 11, Liu as modified teaches the method of claim 10, further comprising:
storing the user input in a database (see Hamm paragraphs [0022], [0030], and [0051]); and
retrieving the user input in response to at least one of the following:
initiation of an update to the Bayesian model (see Hamm paragraphs [0022], [0030], and [0051]); or
initiation of an update to the PELT model (see Hamm paragraphs [0022], [0030], and [0051]).
As to claim 12, Liu as modified teaches a computing system, comprising:
at least one hardware processor (see Liu 11:54-12:36);
at least one memory( coupled to the at least one hardware processor (see Liu 11:54-12:36);
a stored Bayesian model and a stored Pruned Exact Linear Time (PELT) model… (see Liu 3:28-32 and 3:36-41); and
one or more non-transitory computer-readable media having stored therein computer-executable instructions that, when executed by the computing system, cause the computing system to perform operations implementing a combined change-point analyzer for detecting performance regressions in a database system (see Liu 22:16-24. It is noted that the element of “for detecting performance regressions in a database system” is an intended use and has no patentable weight), the operations comprising:
receiving time series data representing performance test results for the database system (see Liu 3:28-41 and the rejection of claim 1);
determining a set of candidate change points comprising a plurality of pre-change points in the time series data by applying the Bayesian model to the time series data (see Liu 3:28-41, 15:8-16, 19:28-24, and 21:33-45 and the rejection of claim 1);
determining one or more change points from among the set of candidate change points by applying the PELT model to the time series data and the plurality of pre-change points (see Liu 3:28-41, 15:8-16, 19:28-24, and 21:33-45 and the rejection of claim 1); and
outputting the one or more change points (see Liu 19:28-42) …
Liu does not teach:
a stored Pruned Exact Linear Time (PELT) model that incorporates a Radial Basis Function (RBF) model;
outputting the one or more change points to a benchmark monitor, wherein the benchmark monitor is configured to receive the one or more change points as inputs and capture one or more performance metrics for performance regression tests;
receiving user input generated via a manual verification process for at least one of the one or more change points; and
adjusting, based on the user input, at least one of a prior distribution of the Bayesian model or a penalty value parameter of the PELT model.
Kaçar teaches a stored Pruned Exact Linear Time (PELT) model that incorporates a Radial Basis Function (RBF) model (see Kaçar page 904, Section III A. The cost function may use a Radial Basis Function).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Liu by the teachings of Kaçar because both references are directed towards statistical analysis of data for the identification of change points using PELT and because Kaçar provides to Liu additional considerations to incorporate into the PELT algorithm that will further minimize costs.
Hamm teaches:
outputting the one or more change points to a benchmark monitor, wherein the benchmark monitor is configured to receive the one or more change points as inputs and capture one or more performance metrics for performance regression tests (see Hamm paragraphs [0022], [0030], and [0051]);
receiving user input generated via a manual verification process for at least one of the one or more change points (see Hamm paragraphs [0022], [0030], and [0051]); and
adjusting, based on the user input, at least one of a prior distribution of the Bayesian model or a penalty value parameter of the PELT model (see Hamm paragraphs [0016], [0022], [0030], and [0051]).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Liu by the teachings of Hamm because both references are directed towards statistical analysis of data for the identification of change points and because Hamm provides to Liu additional input that will allow a model to be improved over time, increasing the utility to a user and the accuracy of a model.
As to claim 13, Liu as modified teaches the system of claim 12, further comprising a pre-processing module comprising computer-executable instructions that, when executed by the computing system, cause the computing system to perform operations comprising:
receiving raw time series data (see Liu 2:63-3:5 and 3:16-17);
pre-processing the raw time series data, wherein the pre-processing of the raw time series data comprises reducing noise in the raw time series data and removing one or more outlier data points from the raw time series data (see Liu 14:60-15:7); and
outputting the pre-processed raw time series data to the combined change-point analyzer as the time series data (see Liu 14:60-15:7).
As to claim 14, Liu as modified by Hamm teaches the system of claim 12, wherein the operations further comprise:
adjusting one or more parameters of the RBF model based on the user input (see Hamm paragraphs [0022], [0030], and [0051]. Hamm shows wherein users may input feedback regarding the identification of changepoints. This feedback may then be used to retrain the models used by the system. As noted above, Kaçar page 904, Section III A teaches an RBF model).
As to claim 15, Liu as modified by Hamm teaches the system of claim 14, wherein the user input comprises at least one of the following for a given change point of the one or more change points:
an indication to designate the given change point as a confirmed change point (see Hamm paragraphs [0022], [0030], and [0051]);
an indication to remove the given change point;
an indication to designate the given change point as a pending change point for further review (see Hamm paragraphs [0022], [0030], and [0051]); or
an indication to adjust a position of the given change point and designate the given change point as a modified change point (see Hamm paragraphs [0022], [0030], and [0051]).
As to claim 16, Liu as modified by Hamm teaches the system of claim 14 further comprising a database, wherein the operations further comprise:
storing the user input in the database (see Hamm paragraphs [0022], [0030], and [0051]); and
retrieving the user input in response to at least one of the following:
initiation of an update to the Bayesian model (see Hamm paragraphs [0022], [0030], and [0051]);
initiation of an update to the PELT model (see Hamm paragraphs [0022], [0030], and [0051]); or
initiation of an update to the RBF model (see Hamm paragraphs [0022], [0030], and [0051]).
As to claim 17, Liu as modified teaches the or more non-transitory computer-readable media comprising computer-executable instructions that, when executed by a computing system, cause the computing system to perform operations implementing a combined change-point analyzer for detecting performance regressions in a database system, the operations comprising:
receiving time series data representing performance test results for the database system (see Liu 2:63-3:5 and 3:16-17);
determining a set of candidate change points comprising a plurality of pre-change points in the time series data by applying a Bayesian model to the time series data (see Liu 3:28-32);
determining one or more change points from among the set of candidate change points by applying a Pruned Exact Linear Time (PELT) model to the time series data and the plurality of pre-change points (see 3:36-41, 15:8-16, 19:28-42, 21:33-45),
…
outputting the one or more change points (see Liu 19:28-42);
receiving user input regarding the one or more change points (see Liu 8:66-9:7); and
Liu does not teach:
wherein applying the PELT model to the time series data and the plurality of pre-change points comprises:
determining respective costs of the plurality of pre-change points using a penalized cost function; and
minimizing the penalized cost function over the time series data;
outputting the one or more change points to a benchmark monitor, wherein the benchmark monitor is configured to receive the one or more change points as inputs and capture one or more performance metrics for performance regression tests;
receiving user input generated via a manual verification process for at least one of the one or more change points; and
adjusting, based on the user input, at least one of a prior distribution of the Bayesian model or a penalty value parameter of the PELT model.
Kaçar teaches:
wherein applying the PELT model to the time series data and the plurality of pre-change points comprises:
determining respective costs of the plurality of pre-change points using a penalized cost function (see Kaçar pages 903-904, Section III and III A); and
minimizing the penalized cost function over the time series data (see Kaçar pages 903-904, Section III and III A);
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Liu by the teachings of Kaçar because both references are directed towards statistical analysis of data for the identification of change points using PELT and because Kaçar provides to Liu additional considerations to incorporate into the PELT algorithm that will further minimize costs.
Hamm teaches:
outputting the one or more change points to a benchmark monitor, wherein the benchmark monitor is configured to receive the one or more change points as inputs and capture one or more performance metrics for performance regression tests (see Hamm paragraphs [0022], [0030], and [0051]);
receiving user input generated via a manual verification process for at least one of the one or more change points (see Hamm paragraphs [0016], [0022], [0030], and [0051]); and
adjusting, based on the user input, at least one of a prior distribution of the Bayesian model or a penalty value parameter of the PELT model (see Hamm paragraphs [0016], [0022], [0030], and [0051]. Hamm shows wherein users may input feedback regarding the identification of changepoints. This feedback may then be used to retrain the model).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Liu by the teachings of Hamm because both references are directed towards statistical analysis of data for the identification of change points and because Hamm provides to Liu additional input that will allow a model to be improved over time, increasing the utility to a user and the accuracy of a model.
.
As to claim 18, Liu as modified by Hamm teaches the computer-readable media of claim 17, wherein the user input comprises one of the following for a given change point of the one or more change points:
an indication to designate the given change point as a confirmed change point (see Hamm paragraphs [0022], [0030], and [0051]);
an indication to remove the given change point;
an indication to designate the given change point as a pending change point for further review (see Hamm paragraphs [0022], [0030], and [0051]); or
an indication to adjust a position of the given change point and designate the given change point as a modified change point (see Hamm paragraphs [0022], [0030], and [0051]).
As to claim 19, Liu as modified by Hamm teaches the computer-readable media of claim 17, wherein the operations further comprise: storing the user input in a database; and
retrieving the user input in response to at least one of the following:
initiation of an update to the Bayesian model (see Hamm paragraphs [0022], [0030], and [0051]); or
initiation of an update to the PELT model (see Hamm paragraphs [0022], [0030], and [0051]).
As to claim 20, Liu as modified by Kaçar teaches the computer-readable media of claim 17, wherein the PELT model incorporates a Radial Basis Function (RBF) model (see Kaçar page 904, Section III A. The cost function may use a Radial Basis Function), and
wherein the operations further comprise:
adjusting one or more parameters of the RBF model based on the user input (see Hamm paragraphs [0022], [0030], and [0051]).
Claims 3-4 are rejected under 35 U.S.C. 103 as being unpatentable over Liu et al. (US Patent 12,306,811) in view of Kaçar et al. (“Automatic Segmentation of Time Series Data with PELT Algorithm for Predictive Maintenance in the Flat Steel Industry”), in view of Hamm (US Pre-Grant Publication 2025/0053823), and further in view of Jhuang et al. (US Pre-Grant Publication 2024/0215926).
As to claim 3, Liu as modified teaches the method of claim 2,
wherein positions of the plurality of pre-change points are specified in a list (see Liu 19:28-42 and 21:33-45), and
Liu does not teach wherein applying the Bayesian model to the time series data further comprises:
comparing the respective Bayes factors for the positions to a predefined threshold;
identifying a subset of the positions based on the comparison, wherein the positions in the subset have respective Bayes factors that exceed the predefined threshold;
designating the positions in the subset as pre-change points; and
adding the positions in the subset to the list.
Jhuang teaches:
wherein applying the Bayesian model to the time series data further comprises:
comparing the respective Bayes factors for the positions to a predefined threshold (see Jhuang paragraphs [0060]-[0062]. A sensitivity threshold is set to which measurements are compared);
identifying a subset of the positions based on the comparison, wherein the positions in the subset have respective Bayes factors that exceed the predefined threshold (see Jhuang paragraphs [0060]-[0062]. Measurements that exceed the threshold may be identified as a change point);
designating the positions in the subset as pre-change points (see Jhuang paragraphs [0060]-[0062]); and
adding the positions in the subset to the list (see Jhuang paragraphs [0060]-[0062]).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Liu by the teachings of Jhuang because both references are directed towards statistical analysis of data for the identification of change points using Bayesian means and because Jhuang provides to Liu additional considerations to incorporate into the Bayesian analysis to ensure accurate results.
As to claim 4, Liu as modified teaches the teaches the method of claim 3, wherein the predefined threshold is a customizable log odds threshold (see Jhuang paragraphs [0060]-[0063]).
Claim 21 is rejected under 35 U.S.C. 103 as being unpatentable over Liu et al. (US Patent 12,306,811) in view of Kaçar et al. (“Automatic Segmentation of Time Series Data with PELT Algorithm for Predictive Maintenance in the Flat Steel Industry”), view of Hamm (US Pre-Grant Publication 2025/0053823), further in view of Glickman (US Pre-Grant Publication 2012/0209795).
As to claim 21, Liu as modified teaches the computer-readable media of claim 20.
Liu as modified does not teach wherein adjusting the one or more parameters of the RBF model based on the user input comprises optimizing kernel parameters of the RBF model.
Glickman teaches wherein adjusting the one or more parameters of the RBF model based on the user input comprises optimizing kernel parameters of the RBF model (see paragraph [0219]. Optimizing kernel parameters of the RBF model is known in the art according to Glickman).
It would have been obvious to one of ordinary skill in the art before the earliest filing date of the invention to have modified Liu by the teachings of Glickman because both references are directed to analyzing and improving statistical analysis and because Glickman shows that optimizing kernel parameters was “known in the art” as of the filing date of Glickman (23 May 2011).
Response to Arguments
Applicant's arguments filed 6 March 2026 have been fully considered but they are not persuasive.
Response to Rejections under 35 USC 101
Applicant argues that “These amendments further clarify that the claimed change point determination is applied in the context of performance regression testing in a database system. In particular, the claimed techniques do not merely determine one or more change points in a vacuum. Rather, the claimed techniques output the determined one or more change points to a benchmark monitor configured to receive change points as inputs and capture performance metrics for performance regression tests. The claimed techniques also adjust specific parameters of the Bayesian model or PELT model based on user input generated via a manual verification process for at least one of the one or more change points. Thus, the amended claims further tie the recited change point determination to a practical workflow for performance regression testing in a database system.”
In response to this argument, it is noted that testing of data is a data analysis and thus a mental process. Applicant has not shown a practical application to the results of this test, but rather only shown that the claims support testing, or analyzing, data. Because Applicant has not shown with citations to the specification that the claimed subject matter improves the processing of a computer or provides an improvement to computing technology over previous methods of testing, Applicant’s argument is unpersuasive. It is noted that no result of the claimed testing is used to affect the operation of a computer system in any way that provides an improvement to the computer system.
Applicant argues that “As discussed above, the amended claims use the determined change points in the context of database performance regression testing, including by outputting the change points to a benchmark monitor for performance regression tests and adjusting specific Bayesian model or PELT model parameters based on user input generated via manual verification of at least one of the change points. Accordingly, the claims are directed to detecting performance regressions in a database system, not merely to data analysis used for additional data analysis.”
In response to this argument, it appears as if “performance regression tests” are data analyses. A data analysis is a mental process step. A claim directed towards mental process steps, without integrating the mental process into a practical application or providing significantly more than the mental process, is patent ineligible.
Applicant has not shown how the results of the test are used in the claims to improve a computing system.
Applicant argues that “Further, the Action's characterization of the claims as reciting a mental process is inconsistent with the amended claim language. As amended, the claims require a specific computer-implemented pipeline for processing time series data representing database performance test results using Bayesian and PELT models, outputting detected change points to a benchmark monitor for performance regression tests, and adjusting specific Bayesian model or PELT model parameters based on user input generated via manual verification of at least one of the change points. These operations are not reasonably characterized as steps practically performed in the human mind or with pen and paper.”
In response to this argument, it is noted that the mental process steps of the claims are largely directed towards data analysis steps, accompanied by data input and output steps. These data analysis steps may be performed by a human being with a generic computer.
As noted in MPEP 2106.04(a)(2) III C, “claims can recite a mental process even if they are claimed as being performed on a computer.
The Supreme Court recognized this in Benson, determining that a mathematical algorithm for converting binary coded decimal to pure binary within a computer’s shift register was an abstract idea. The Court concluded that the algorithm could be performed purely mentally even though the claimed procedures "can be carried out in existing computers long in use, no new machinery being necessary." 409 U.S at 67, 175 USPQ at 675. See also Mortgage Grader, 811 F.3d at 1324, 117 USPQ2d at 1699 (concluding that concept of "anonymous loan shopping" recited in a computer system claim is an abstract idea because it could be "performed by humans without a computer").”
MPEP 2106.04(a)(2) III C 1-3 further elaborate on the idea that a claim may still be directed towards an abstract idea despite the use of a generic machine. Thus, though the claims may not be performed solely in a human mind, the determining, applying, minimizing, capturing, and adjusting steps may be performed by a human with a generic computer.
Applicant argues that “Furthermore, even assuming arguendo that the claims recite an abstract idea, which Applicant disputes, the amended claims integrate any such abstract idea into a practical application. For example, the claims do not merely determine one or more change points; rather, the claims use the determined change points in the context of database performance regression testing and then adjust particular parameters of the Bayesian model or PELT model based on user input generated via a manual verification process. These limitations impose meaningful limits on any alleged abstract idea and apply the recited change point determination in the technological field of database performance regression testing.”
As noted above, it is noted that the “technological field of database performance regression testing” appears to be data analysis. It is noted that the results of this analysis are not used in any manner beyond adjusting a model. Applicant has not shown, with citations to the specification, how the claimed steps improve the processing of a computer or provide significantly more than the abstract idea.
Applicant argues that “The present amendments also further support a finding of eligibility under Step 2B. The claims recite a specific ordered combination of features, including determining a set of candidate change points comprising a plurality of pre-change points using a Bayesian model, determining one or more change points from among that set using a PELT model, outputting the determined one or more change points to a benchmark monitor configured to receive the one or more change points as inputs and capture performance metrics for performance regression tests, receiving user input generated via a manual verification process for at least one of the one or more change points, and adjusting a prior distribution of the Bayesian model or a penalty value parameter of the PELT model based on that user input. This ordered combination is not merely a generic instruction to analyze data using a computer.”
In response to this argument, Examiner requests that Applicant cite to a portion of the specification that shows how the claimed steps result in an improvement to the processing of a computer or provide an improvement to a technological field beyond merely an improved method of data analysis.
Response to Rejections under 35 USC 103
Applicant argues that “Applicant respectfully traverses the § 103 rejections at least for the reasons set forth in the previous response. In the previous response, Applicant explained why Liu and Kacar fail to disclose or suggest determining a set of candidate change points comprising a plurality of pre- change points by applying a Bayesian model, and then determining one or more change points from among that set by applying a PELT model. Applicant maintains those arguments.”
In response to this argument, those arguments were addressed in the previous response. Examiner maintains those positions.
Applicant argues that “Applicant also respectfully disagrees with the Action's assertion that considering multiple change points instead of a single change point is a mere duplication of parts. The claims do not rely on patentability merely because more than one change point may be considered. Rather, as discussed in the previous response, the claims require determining a set of candidate change points comprising a plurality of pre-change points by applying a Bayesian model, and then determining one or more change points from among that set by applying a PELT model to the time series data and the plurality of pre-change points. Thus, the claimed features define a particular relationship between the Bayesian model output and the subsequent PELT model analysis. This is not a mere duplication of a known part, but a specific ordered arrangement for using candidate change points determined by applying a Bayesian model in a subsequent PELT- based change point determination.”
In response to this argument, Liu shows how output from a first model may be used as input to a second model to confirm statistical analysis. Initial, or candidate, changepoint data may be analyzed by a second model that outputs a verification status of the changepoint data (see Liu 19:28-42 and 21:33-45). It is also noted that the models are used to detect multiple false-positive changepoints (see Liu 1:47-57 and 11:16-18). Thus, multiple changepoints as candidate change points may be considered by Liu. Or, in other words, multiple sets of changepoints may be considered.
Examiner is not relying on the idea that duplication of parts is obvious for the idea of “duplicating” a changepoint in a first model to be analyzed in a second model, as Applicant appears to be arguing. Instead, Examiner is relying on the idea of duplicating parts being obvious for the idea of analyzing multiple candidates, or “duplicating” the changepoint being analyzed by both models of Liu such that it would be obvious that multiple such changepoints would be analyzed by Liu.
Applicant argues that “Liu and Kacar fail to disclose or suggest the newly added limitations. The Action relies on Liu for change point detection and false-positive analysis, and relies on Kacar for PELT- related features. At most, the cited portions of Liu and Kacar describe identifying or analyzing change points. They do not disclose or suggest the outputting one or more change points to a benchmark monitor configured to receive change points as inputs and capture performance metrics for performance regression tests.”
In response to this argument, previously cited reference Hamm is relied upon to teach these limitations.
Applicant argues that “Whether or not this is true, Applicant submits that Hamm fails to disclose or suggest the more specific limitations now recited in the amended independent claims. As amended, the independent claims do not merely require generic user feedback or generic retraining. Rather, the independent claims require user input generated via a manual verification process and adjustment, based on that user input, of a Bayesian model prior distribution or PELT model penalty value parameter. The Action has not shown that Hamm discloses or suggests these features. Generic feedback regarding identification of change points, even if used to retrain a model, does not disclose or suggest the claimed adjustment, based on user input generated via a manual verification process, of a Bayesian model prior distribution or PELT model penalty value parameter.”
In response to this, it is noted that “Generic feedback regarding identification of change points, even if used to retrain a model” is, under a broadest reasonable interpretation, “user input generated via a manual verification process for at least one of the one or more change points.” Applicant does not describe any particular format or criteria that the “user input” must take, beyond that it related somehow to “at least one of the one or more change points.” It is additionally noted that the feedback provided does adjust bias metric , which may be used to change update Bayesian models 114 and 116 in Hamm (see paragraphs [0016], [0022], and [0051]).
Applicant argues that “Nor has the Action provided a sufficient reason why a person of ordinary skill would have modified the combination of Liu and Kacar to arrive at the claimed feedback-related features. The Action's rationale for combining Hamm with Liu and Kacar is that feedback may allow a model to be improved over time. But that rationale is stated at a high level of generality, and does not explain why or how a person of ordinary skill would have used manually-verified change point information to adjust a Bayesian prior distribution or a PELT penalty value parameter in the Liu/Kacar system. At most, the Action identifies a general desirability of using feedback to improve a model, which is insufficient to supply the particular claimed implementation.”
In response to this argument, it is noted that the references teach the claimed subject matter to the extent claimed for the reasons provided in the rejection above. It is noted that improving a model is sufficient reason for a person of ordinary skill to have used manually-verified change point information to adjust a model.
Applicant is reminded that the claim limitation “PELT penalty value parameter” is claimed as an alternative and not required by the claim.
Applicant argues that “The asserted combination therefore fails to disclose or suggest amended claim 1. Liu and Kacar fail to disclose or suggest the claimed two-stage Bayesian/PELT change point determination for the reasons discussed in the previous response. In addition, Liu, Kacar, and Hamm fail to disclose or suggest the newly added limitations relating to benchmark monitor use and adjustment of a Bayesian prior distribution or PELT penalty value parameter based on user input generated via a manual verification process for at least one of the one or more change points.”
In response to this argument, it is noted that the amended subject matter of the independent claims is taught by a combination of Liu, Kaçar, and Hamm for the reasons provided in the office action above.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CHARLES D ADAMS whose telephone number is (571)272-3938. The examiner can normally be reached M-F, 9-5:30 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, Aleksandr Kerzhner can be reached at 5712701760. 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.
/CHARLES D ADAMS/ Primary Examiner, Art Unit 2165
Prosecution Insights