DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 5/15/2026 has been entered.
Response to Amendment
The Amendment filed 5/15/2026 has been entered. Claims 1-2, 8-9, and 15 have been amended. Claims 1-20 are pending in the application.
Drawings
The drawings are objected to as failing to comply with 37 CFR 1.84(p)(5) because they include the following reference character(s) not mentioned in the description: reference character 216 (FIG. 2). Corrected drawing sheets in compliance with 37 CFR 1.121(d), or amendment to the specification to add the reference character(s) in the description in compliance with 37 CFR 1.121(b), are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either "Replacement Sheet" or "New Sheet" pursuant to 37 CFR 1.121(d). The objection to the drawings will not be held in abeyance.
Specification
The disclosure is objected to because of the following informalities: paragraph [0038] refers to the temporal aggregation unit as "a temporal aggregation unit 122," whereas the same temporal aggregation unit is designated by reference character 104 elsewhere in the disclosure, including at paragraphs [0027] and [0030] and in FIG. 1. The inconsistent numbering of this element should be corrected so that the temporal aggregation unit is consistently designated by reference character 104. Appropriate correction is required.
Claim Objections
Claims 8 and 16 are objected to because of the following informalities:
In claim 8, the clause "receive a request for a first forecasted time step value and a second forecasted time step value" and the clause "determine a first number of time steps of a forecasting horizon associated with the second forecasted time step value" each conclude with a colon; each colon should be a semicolon, consistent with the punctuation used to separate the other clauses in the recited series of instructions.
In claim 16, the recitation "a first machine learning model to the determine the first forecasted time step value" includes an extraneous word "the" preceding "determine"; the limitation should read "a first machine learning model to determine the first forecasted time step value."
Appropriate correction is required.
Claim Rejections - 35 U.S.C. 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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claim 1
Step 1: Claim 1 recites a computer-implemented method and is therefore directed to a process, one of the four statutory categories of invention.
Step 2A, Prong One: Claim 1 recites a judicial exception.
The limitation generating ... a temporally aggregated time series by summing a set of sequential time step values from the time series, wherein a quantity of time step values in the set of time step values is equal to the first number of time steps of the forecasting horizon sets forth a mathematical calculation.
The limitation calculating ... a first set of input values from the time series and a second set of input values from the temporally aggregated time series sets forth a further mathematical calculation.
The limitations determining ... the first forecasted time step value and determining ... the second forecasted time step value ... independent of the determining of the first forecasted time step value set forth an evaluation, namely a forecasting or estimation judgment of the kind that can be performed mentally or with pen and paper. These limitations recite mathematical concepts and a mental process (MPEP 2106.04(a)(2)).
Step 2A, Prong Two: The judicial exception is not integrated into a practical application. Beyond the recited exception, claim 1 adds only (i) receiving ... a time series comprising time step values and receiving ... a request for a first forecasted time step value and a second forecasted time step value, and (ii) a generic computing device that performs the steps. The receiving steps are insignificant extra-solution data gathering (MPEP 2106.05(g)). The computing device is recited at a high level of generality and merely represents generic computer machinery performing in its ordinary capacity to implement the underlying judicial exception (MPEP 2106.05(f)).
The benefit on which the applicant relies - reducing the cross-horizon error described at specification paragraphs [0025]-[0026] - is an improvement in the accuracy of the forecast itself, which is the output of the abstract idea, and is not an improvement to the functioning of a computer or to any other technology or technical field (MPEP 2106.05(a); Electric Power Group v. Alstom; SAP America v. InvestPic; Recentive Analytics v. Fox). The limitations that produce that benefit - the summing that generates the temporally aggregated series, the calculating of the input values, and the determining of each forecasted value independent of the other - are themselves the recited mathematical concepts and evaluation; the judicial exception alone cannot provide the improvement (MPEP 2106.05(a); Ex Parte Desjardins). The forecasted values are not applied to control, transform, or otherwise act upon any device or process; the claim ends at producing numbers. Considered individually and as a whole, the additional elements do not integrate the exception into a practical application.
Step 2B: The additional elements, individually and as an ordered combination, do not amount to significantly more than the judicial exception. Receiving the time series and the forecast requests is receiving and transmitting data over a network, a well-understood, routine, and conventional computer function (Symantec; TLI Communications; buySAFE; OIP Technologies). The summing and calculating are repetitive mathematical computations performed by a generic computer (Flook; Bancorp). The computing device is a generic computer component performing its ordinary functions (Alice). Considered as an ordered combination, these elements add nothing beyond the sum of their parts. Claim 1 does not amount to significantly more and is ineligible.
Claims 2-7
Claims 2-7 depend from claim 1 and are rejected under 35 U.S.C. 101 for the same reasons. Each incorporates the abstract idea of claim 1 and adds no element that integrates the exception or amounts to significantly more.
Claim 2 adds that the computing device implements a first machine learning model and a second machine learning model to determine the forecasted values; these are generic model components recited as mere instructions to implement the abstract idea, equivalent to adding the words apply it (MPEP 2106.05(f)).
Claim 3 adds that both models implement a same forecasting technique, and claim 4 adds that the technique is an autoregressive moving average technique; naming a conventional forecasting technique (see specification [0032] and [0045], describing a suite of forecasting techniques including ARMA and ARIMA) adds only further mathematical and evaluation subject matter and no inventive concept.
Claim 5 adds that the input values comprise a trend, a seasonality, an autocorrelation, a nonlinearity, or a heterogeneity of the time series, which are further statistical features of the same mathematical analysis.
Claim 6 adds discarding ... an oldest time step value, an insignificant extra-solution data-management step (MPEP 2106.05(g)).
Claim 7 adds training the first machine learning model via the forecasting technique, a mere instruction to implement the abstract idea on a generic model, equivalent to apply it (MPEP 2106.05(f)).
None of claims 2-7 recites an additional element that integrates the exception or supplies significantly more.
Claims 8-14
Step 1: Claim 8 recites a cloud infrastructure node comprising a processor and a computer-readable medium and is directed to a machine.
Step 2A, Prong One: Claims 8-14 recite the same abstract ideas as in claims 1-7, respectively.
Step 2A, Prong Two: The judicial exceptions are not integrated into a practical application; the analysis at this step mirrors that of claims 1-7, respectively. The additional cloud infrastructure node, processor, and computer-readable medium are generic computer components recited at a high level of generality and merely represent generic computer machinery that cannot integrate the exception (MPEP 2106.05(f)).
Step 2B: These claims do not contain significantly more; the analysis at this step mirrors that of claims 1-7, respectively.
Claims 15-20
Step 1: Claim 15 recites a non-transitory computer-readable medium and is directed to an article of manufacture; because the medium is expressly non-transitory, no transitory-signal issue arises.
Step 2A, Prong One: Claims 15-20 recite the same abstract ideas as in claims 1-6, respectively (claim 15 as claim 1; claims 16-20 as claims 2-6).
Step 2A, Prong Two: The judicial exceptions are not integrated into a practical application; the analysis at this step mirrors that of claims 1-6, respectively. The only difference is the non-transitory computer-readable-medium category, which is generic computer equipment and cannot integrate the exception (MPEP 2106.05(f)).
Step 2B: These claims do not contain significantly more; the analysis at this step mirrors that of claims 1-6, respectively.
Claim Rejections - 35 U.S.C. 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 1-3, 6-10, 13-17, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Bansal et al. (hereinafter Bansal), US 2017/0220939 A1, in view of Rostami-Tabar et al. (hereinafter Rostami-Tabar) “To aggregate or not to aggregate: Forecasting of finite autocorrelated demand“ (2021), and further in view of Ben Taieb et al. (hereinafter Ben Taieb) “A review and comparison of strategies for multi-step ahead time series forecasting based on NN5 forecasting competition” (2011).
Regarding independent claim 1, Bansal teaches a computer-implemented method comprising (Bansal: Abstract "A multi-horizon predictor system that predicts a future parameter value for multiple horizons based on time-series data of the parameter, external data, and machine-learning"; the multi-horizon predictor is implemented on a computing system having at least one hardware processing unit, so that the recited acts are performed by a computer):
receiving, by a computing device, a time series comprising time step values (Bansal: [0033], FIG. 2 "The time series data splitter 210 receives time series data 201 of a parameter and an identified horizon 202"; the time series data 201 (a time series) is received by the time series data splitter 210 of the computing system (a computing device) and comprises successive parameter values over time (time step values));
receiving, by the computing device, a request for a first forecasted time step value and a second forecasted time step value (Bansal: [0042] "The prediction trigger 240 causes the multi-horizon prediction model to redo the multi-horizon prediction", claim 1 "to generate a multi-horizon prediction result", [0038] "The multi-horizon predictor 230 may gather each result of the final prediction models for each of the multiple horizons to thereby generate a multi-horizon prediction"; the prediction trigger 240 receives a request that causes the multi-horizon prediction result comprising predictions for multiple horizons to be generated, a prediction for a nearer horizon being the first forecasted time step value and a prediction for a longer horizon being the second forecasted time step value);
determining, by the computing device, a first number of time steps of a forecasting horizon associated with the second forecasted time step value (Bansal: [0033], FIG. 2 "receives time series data 201 of a parameter and an identified horizon 202", [0034] "the time period 301 and horizons may be on the order of years, quarters, months, weeks, days, hours, and even minutes"; the identified horizon 202 (a forecasting horizon) for the longer horizon prediction spans a determined number of time increments (a first number of time steps) associated with the second forecasted time step value, e.g. horizons on the order of years, quarters, months, weeks, days, hours, and even minutes);
calculating, by the computing device, a first set of input values from the time series (Bansal: [0005], "fitting an initial prediction model to the parameter using the training data thereby using machine learning"; the training data drawn from the time series data 201 is calculated and supplied to the initial prediction model (a first set of input values from the time series)); and
determining, by the computing device, the first forecasted time step value using the first set of input values (Bansal: [0006] "a different final prediction model used for each of at least some of the multiple time horizons"; the final prediction model for the nearer horizon generates that horizon's prediction (the first forecasted time step value) from the training data inputs (the first set of input values)).
Bansal does not expressly teach generating, by the computing device, a temporally aggregated time series by summing a set of sequential time step values from the time series, wherein a quantity of time step values in the set of time step values is equal to the first number of time steps of the forecasting horizon; and a second set of input values from the temporally aggregated time series, the first set of input values and the second set of input values being based at least in part on a same set of input features; and determining, by the computing device, the second forecasted time step value using the temporally aggregated time series.
However, Rostami-Tabar teaches generating, by the computing device, a temporally aggregated time series by summing a set of sequential time step values from the time series, wherein a quantity of time step values in the set of time step values is equal to the first number of time steps of the forecasting horizon (Rostami-Tabar: page 2 "the time series are divided into consecutive non-overlapping buckets of time where the length of the time bucket equals the aggregation level. The aggregate demand is created by summing up the values inside each bucket", page 6 "The aggregation level is conveniently chosen to match the forecast horizon"; a non-overlapping temporally aggregated series (a temporally aggregated time series) is created by summing the sequential observations within each bucket (summing a set of sequential time step values), the bucket length (a quantity of time step values in the set of time step values) equaling the aggregation level, which is chosen to match the forecast horizon (equal to the first number of time steps of the forecasting horizon)); and a second set of input values from the temporally aggregated time series, the first set of input values and the second set of input values being based at least in part on a same set of input features (Rostami-Tabar: page 4 "We analyse the performance of the three approaches: non-aggregation, non-overlapping aggregation, and overlapping aggregation", page 3, Figure 1, page 5 "using SES as forecasting method"; the same single exponential smoothing SES forecasting inputs (a same set of input features) are computed both from the original series (the first set of input values) and from the non-overlapping aggregated series (a second set of input values from the temporally aggregated time series)); and determining, by the computing device, the second forecasted time step value using the temporally aggregated time series (Rostami-Tabar: page 6 "Our analysis refers to the aggregated series", page 12 "Using SES, the forecast of non-overlapping aggregated series in period T"; the longer horizon forecast (the second forecasted time step value) is produced from the non-overlapping aggregated series (using the temporally aggregated time series)).
Because Bansal and Rostami-Tabar are analogous art and within the same field of endeavor, specifically time-series forecasting, and address the same problem-solving area of accurately forecasting future values over a forecasting horizon, accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to combine the non-overlapping temporal aggregation of Rostami-Tabar, in which the aggregation level is chosen to match the forecast horizon, with the multi-horizon forecasting framework of Bansal, with a reasonable expectation of success, so as to teach generating, by the computing device, a temporally aggregated time series by summing a set of sequential time step values from the time series, wherein a quantity of time step values in the set of time step values is equal to the first number of time steps of the forecasting horizon together with the calculation and use of the second set of input values from that temporally aggregated time series to teach calculating, by the computing device, a first set of input values from the time series and a second set of input values from the temporally aggregated time series, the first set of input values and the second set of input values being based at least in part on a same set of input features; and determining, by the computing device, the second forecasted time step value using the temporally aggregated time series. This modification would have been motivated by the desire to improve longer-horizon forecast accuracy through temporal aggregation matched to the horizon (Rostami-Tabar: page 3 "the aggregation approach can improve forecast accuracy as compared to the non-aggregation approach").
Bansal and Rostami-Tabar do not expressly teach determining the second forecasted time step value independent of the determining of the first forecasted time step value.
However, Ben Taieb teaches determining a forecast for a horizon independent of the determining of the first forecasted time step value (Ben Taieb: page 8 "The Direct (also called Independent) strategy ... consists of forecasting each horizon independently from the others. In other terms, H models fh are learned (one for each horizon) from the time series…the H models are learned independently inducing a conditional independence of the H forecasts"; under the Direct, or Independent, strategy each horizon's forecast is produced by its own model learned independently of the other horizons' models, so that the longer-horizon forecast is determined independent of the determining of the first forecasted time step value).
Because Bansal, in view of Rostami-Tabar, and Ben Taieb are analogous art and within the same field of endeavor, specifically multi-step-ahead time-series forecasting, and address the same problem-solving area of generating accurate forecasts at multiple horizons, accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to determine the longer-horizon forecast of the Bansal and Rostami-Tabar combination using the Direct (Independent) strategy of Ben Taieb, with a reasonable expectation of success, thereby teaching determining, by the computing device, the second forecasted time step value using the temporally aggregated time series independent of the determining of the first forecasted time step value. This modification would have been motivated by the desire to avoid the accumulation of forecast errors that attends recursive multi-step forecasting (Ben Taieb: page 8 "the Direct strategy does not use any approximated values to compute the forecasts (Equation 4), being then immune to the accumulation of errors").
Regarding dependent claim 2, Bansal, in view of Rostami-Tabar and Ben Taieb, teach the method of claim 1, wherein the computing device implements a first machine learning model to determine the first forecasted time step value and a second machine learning model to determine the second forecasted time step value (Bansal: [0006] "a different tuning and thus a different final prediction model used for each of at least some of the multiple time horizons"; the final prediction model for the nearer horizon (a first machine learning model), fitted by machine learning, determines the first forecasted time step value, and the different final prediction model for the longer horizon (a second machine learning model) determines the second forecasted time step value).
Regarding dependent claim 3, Bansal, in view of Rostami-Tabar and Ben Taieb, teach the method of claim 2, wherein both the first machine learning model and the second machine learning model implement a same forecasting technique (Bansal: [0039] "the model tuner 220 keeps the same initial prediction model for each of the prediction models for each horizon of a single multi-horizon prediction"; claim 2 "the multi-horizon predictor keeping the same initial prediction model for each of the plurality of horizons"; keeping the same initial prediction model for each horizon's prediction model causes both the first machine learning model and the second machine learning model to implement a same forecasting technique).
Regarding dependent claim 6, Bansal, in view of Rostami-Tabar and Ben Taieb, teach the method of claim 1, wherein the method further comprises discarding a sixth time step value, and wherein the sixth time step value is an oldest time step value of the time series (Rostami-Tabar: page 2 "At each period, the window is moved one step ahead, so the oldest observation is dropped and the newest is included"; as the aggregation window advances by one step, the oldest observation (a sixth time step value that is an oldest time step value of the time series) is dropped from the series).
Regarding dependent claim 7, Bansal, in view of Rostami-Tabar and Ben Taieb, teach the method of claim 3, wherein the method further comprises training the first machine learning model via the forecasting technique (Bansal: [0005] "fitting an initial prediction model to the parameter using the training data thereby using machine learning"; the initial prediction model (the first machine learning model) is trained by fitting it to the parameter using the training data via the forecasting technique).
Regarding independent claim 8, it is a code infrastructure node claim that is substantially the same as the computer-implemented method of claim 1. Thus, claim 8 is rejected for the same reason as claim 1. In addition, Bansal teaches a cloud infrastructure node, comprising: a processor; and a computer-readable medium including instructions that, when executed by the processor, cause the processor to perform the recited operations (Bansal: [0031] "the invention may be practiced in a cloud computing environment…’cloud computing' is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services)" (a cloud infrastructure node); [0017] "a computing system 100 typically includes at least one hardware processing unit 102 and memory 104" (a processor); claim 20 "A computer program product comprising one or more computer-readable storage media having thereon computer-executable instructions that are structure such that, when executed by one or more processors of a computing system, cause the computing system to perform a method" (a computer-readable medium including instructions that, when executed by the processor, cause the processor to); and [0043] "accurately predict the number of processing nodes, storage nodes, power usage, and so forth, based on pass resource usage" (situating the multi-horizon predictor within a cloud-infrastructure resource-forecasting node)).
Regarding dependent claims 9-10 and 13-14, these are cloud infrastructure node claims that are substantially the same as the computer-implemented method of claims 2-3 and 6-7, respectively. Thus, claims 9-10 and 13-14 are rejected for the same reasons as claims 2-3 and 6-7.
Regarding independent claim 15, it is a non-transitory computer-readable medium claim that is substantially the same as the computer-implemented method of claim 1. Thus, claim 15 is rejected for the same reason as claim 1. In addition, Bansal teaches a non-transitory computer-readable medium having stored thereon a sequence of instructions which, when executed, causes a processor to perform operations comprising the recited operations (Bansal: [0025] "Computer-readable storage media includes RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other physical and tangible storage medium which can be used to store desired program code means in the form of computer-executable instructions" (a non-transitory computer-readable medium having stored thereon a sequence of instructions); claim 20 "when executed by one or more processors of a computing system, cause the computing system to perform a method" (which, when executed, causes a processor to perform operations)).
Regarding dependent claims 16-17 and 20, these are non-transitory computer-readable medium claims that are substantially the same as the computer-implemented method of claims 2-3 and 6, respectively. Thus, claims 16-17 and 20 are rejected for the same reasons as claims 2-3 and 6.
Claims 4, 11, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Bansal, in view of Rostami-Tabar and Ben Taieb, as applied in the rejections of claims 1, 8, and 15, respectively above, and further in view of Petropoulos et al. (hereinafter Petropoulos) “Forecasting: theory and practice” (Jan 2022).
Regarding dependent claim 4, Bansal, in view of Rostami-Tabar and Ben Taieb, teach all the elements of claim 3.
Bansal, Rostami-Tabar, and Ben Taieb do not expressly teach wherein the forecasting technique is an autoregressive moving average technique.
However, Petropoulos teaches wherein the forecasting technique is an autoregressive moving average technique (Petropoulos: page 20, Section 2.3.4, "Time series models that are often used for forecasting are of the autoregressive integrated moving average class (ARIMA - Box et al., 1976)"; the autoregressive integrated moving average (ARIMA) class is an autoregressive moving average technique used to forecast a time series).
Because Bansal, in view of Rostami-Tabar and Ben Taieb, and Petropoulos are analogous art and within the same field of endeavor, specifically time-series forecasting, and address the same problem-solving area of selecting a forecasting technique for generating forecasts, accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to implement the same forecasting technique of the combination as an autoregressive moving average technique as taught by Petropoulos, with a reasonable expectation of success, thereby teaching wherein the forecasting technique is an autoregressive moving average technique. This modification would have been motivated by the desire to employ a well-established and widely used class of forecasting models (Petropoulos: page 20, Section 2.3.4, "Time series models that are often used for forecasting are of the autoregressive integrated moving average class").
Regarding dependent claim 11, it is a cloud infrastructure node claim that is substantially the same as the computer-implemented method of claim 4. Thus, claim 11 is rejected for the same reason as claim 4.
Regarding dependent claim 18, it is a non-transitory computer-readable medium claim that is substantially the same as the computer-implemented method of claim 4. Thus, claim 18 is rejected for the same reason as claim 4.
Claims 5, 12, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Bansal, in view of Rostami-Tabar and Ben Taieb, as applied in the rejections of claims 1, 8, and 15, respectively above, and further in view of Talagala et al. (hereinafter Talagala) “Meta-learning how to forecast time series” (2018).
Regarding dependent claim 5, Bansal, in view of Rostami-Tabar and Ben Taieb, teach all the elements of claim 1.
Bansal, Rostami-Tabar, and Ben Taieb do not expressly teach wherein the first set of input values comprises a trend, a seasonality, an autocorrelation, a nonlinearity, or a heterogeneity of the time series.
However, Talagala teaches wherein the first set of input values comprises a trend, a seasonality, an autocorrelation, a nonlinearity, or a heterogeneity of the time series (Talagala: page 17, Table 2: Features used for selecting a forecast-model, "strength of trend", "strength of monthly seasonality", "first ACF value of the original series", and "nonlinearity"; the FFORMS feature vector computed from each time series are selecting from inputs comprising a strength of trend (a trend), a strength of seasonality (a seasonality), a first autocorrelation-function value (an autocorrelation), and a nonlinearity (a nonlinearity) of the time series).
Because Bansal, in view of Rostami-Tabar and Ben Taieb, and Talagala are analogous art and within the same field of endeavor, specifically feature-based time-series forecasting, and address the same problem-solving area of characterizing a time series for forecasting, accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to compute the first set of input values as the time-series features taught by Talagala, with a reasonable expectation of success, thereby teaching wherein the first set of input values comprises a trend, a seasonality, an autocorrelation, a nonlinearity, or a heterogeneity of the time series. This modification would have been motivated by the desire to select an appropriate forecasting model based on measurable characteristics of the time series (Talagala: page 3, "Clearly, there is a need for a fast and scalable algorithm to automate the process of selecting models with the aim of forecasting").
Regarding dependent claim 12, it is a cloud infrastructure node claim that is substantially the same as the computer-implemented method of claim 5. Thus, claim 12 is rejected for the same reason as claim 5.
Regarding dependent claim 19, it is a non-transitory computer-readable medium claim that is substantially the same as the computer-implemented method of claim 5. Thus, claim 19 is rejected for the same reason as claim 5.
Response to Arguments
Regarding the objection to claim 9, Applicant argues that claim 9 has been amended as suggested by the Examiner and that the objection should be withdrawn (Remarks, page 7). The argument is persuasive. The amendment to claim 9 cures the noted informality, and the objection to claim 9 set forth in the final Office Action is withdrawn. However, the present amendment introduces new informalities in claims 8 and 16, and new objections to those claims, necessitated by amendment, are set forth above.
Regarding the rejection of claims 1-20 under 35 U.S.C. 112(b), Applicant argues that the claims have been amended as suggested by the Examiner (Remarks, page 7). The argument is persuasive. The amendments to claims 1, 8, and 15, including the recitations of summing "a set of sequential time step values" and "wherein a quantity of time step values in the set of time step values is equal to the first number of time steps of the forecasting horizon," cure the indefiniteness identified in the final Office Action, and the amended claims particularly point out and distinctly claim the subject matter regarded as the invention. The rejection of claims 1-20 under 35 U.S.C. 112(b) is withdrawn.
Regarding the rejection of claims 1-20 under 35 U.S.C. 101, Applicant's arguments (Remarks, pages 8-11) have been fully considered but are not persuasive. The rejection is maintained and is restated above as applied to the claims as amended, any changes in its articulation being necessitated by the present amendment, including the newly added limitation "independent of the determining of the first forecasted time step value."
Applicant first argues, citing the August 4, 2025 memorandum "Reminders on Evaluating Subject Matter Eligibility of Claims under 35 U.S.C. 101," that the limitations "determining, by the computing device, a first number of time steps of a forecasting horizon associated with the second forecasted time step value," "determining, by the computing device, the first forecasted time step value," and "determining, by the computing device, the second forecasted time step value using the temporally aggregated time series" cannot practically be performed in the human mind and therefore are not mental processes, such that the claims do not recite a judicial exception (Remarks, pages 8-9).
Examiner respectfully disagrees. This argument is not persuasive, for two reasons. First, the rejection does not rest on the mental-process grouping alone. As set forth above, claims 1-20 recite mathematical concepts: the limitations "generating, by the computing device, a temporally aggregated time series by summing a set of sequential time step values from the time series" and "calculating, by the computing device, a first set of input values from the time series and a second set of input values from the temporally aggregated time series" set forth mathematical calculations in words. Under the very memorandum Applicant invokes, a limitation recites a mathematical concept when it sets forth or describes a calculation using words or mathematical symbols, and "summing" names the calculation performed. Applicant's Remarks do not address the mathematical-concepts grouping or the summing and calculating limitations, and Step 2A Prong One is satisfied on that grouping alone.
Second, the mental-process characterization of the argued "determining" limitations is proper at the level of generality claimed. The memorandum cautions against expanding the mental-process grouping to limitations that cannot practically be performed in the human mind, expressly including limitations that encompass artificial intelligence in that way, such as complex neural-network operations or multidimensional matrix calculations. Independent claims 1, 8, and 15, however, recite no machine learning model, neural network, matrix operation, or any other limitation of that character; machine learning models enter the claim set only in dependent claims 2, 9, and 16, and model training only in claims 7 and 14. As recited, determining "a first number of time steps of a forecasting horizon" is determining a count of time steps, of which the specification's own example is a horizon of two ([0035], [0036]), and determining each forecasted value from a set of input values is, at the claimed level of generality, an evaluation that can practically be performed in the human mind or with pen and paper. The recitation that the steps are performed "by a computing device" does not remove them from the grouping, because a claim that merely uses a computer as a tool to perform a mental process still recites a mental process (MPEP 2106.04(a)(2)(III)(C)). It is further noted that in Ex parte Desjardins (PTAB 2025) (precedential), a claim directed to training a machine learning model was held to recite an abstract idea at Step 2A Prong One, with eligibility resolved at Prong Two; the memorandum does not carry even a machine-learning training claim out of Prong One, much less the present independent claims, which recite no machine learning at all.
Applicant next argues that [0025] and [0026] of the published specification describe a technical problem, namely the carry-over of forecast errors across successive predicted values in conventional multi-step forecasting, and a technical improvement, namely the use of separate forecasting models in combination with temporally aggregated data points, and that the claims reflect the disclosed improvement and integrate any judicial exception into a practical application (Remarks, pages 9-11).
Examiner respectfully disagrees. This argument is not persuasive. The claim language Applicant quotes as reflecting the improvement, namely the generating and summing limitation, the calculating limitation, and the determining limitations (Remarks, pages 10-11), is in each instance the judicial exception itself as identified above: summing and calculating are the recited mathematical calculations, and determining the forecasted values is the recited forecasting evaluation. Per MPEP 2106.05(a), the judicial exception alone cannot provide the improvement, and an asserted improvement that lies entirely within the abstract idea cannot integrate that idea into a practical application. A claimed advance consisting of more accurate mathematical forecasting remains within the realm of abstract ideas. See Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350 (Fed. Cir. 2016); SAP America, Inc. v. InvestPic, LLC, 898 F.3d 1161 (Fed. Cir. 2018); Recentive Analytics, Inc. v. Fox Corp., 134 F.4th 1205 (Fed. Cir. 2025), in which the Federal Circuit held that applying machine learning in its ordinary capacity to produce better predictions in a field of endeavor is abstract absent an improvement to the computer or to the machine-learning mechanism itself.
Moreover, the specific mechanism the specification credits with preventing error carry-over is that "a separate model is employed for each successive predicted value" ([0026]; see also [0050]). Independent claims 1, 8, and 15 do not recite separate models, or any model; the two-model architecture appears only in dependent claims 2, 9, and 16, and amended claim 1 encompasses determining both forecasted values with a single model. Under Ex parte Desjardins and MPEP 2106.04(d)(1), an improvement that is described in the specification but not reflected in the claim cannot integrate the exception. Applicant's observation that, per the memorandum, the claim need not explicitly recite the improvement does not assist, because the claim must still reflect the disclosed improvement, and here the mechanism the specification credits is absent from the independent claims. The asserted benefit, avoiding the carry-over of forecast errors, is in any event an improvement in forecast accuracy, which is the abstract idea's own output, and not an improvement in the functioning of the computing device, the cloud infrastructure node, or any other technology; the claims end at producing forecasted values, and no recited element applies the forecasts to control, transform, or otherwise affect anything technological.
The record art further confirms that the asserted advance was a known forecasting strategy rather than an unconventional technical solution. The limitation added by the present amendment, determining the second forecasted time step value "independent of the determining of the first forecasted time step value," corresponds to the strategy Ben Taieb describes at page 8: "The Direct (also called Independent) strategy ... consists of forecasting each horizon independently from the others," which "does not use any approximated values to compute the forecasts ..., being then immune to the accumulation of errors." Applicant's own paragraph [0025] lists "direct multi-step forecasting" among the "generally accepted" multi-step forecasting techniques. Likewise, temporal aggregation with the aggregation window matched to the forecast horizon was known (Rostami-Tabar: page 6, "The aggregation level is conveniently chosen to match the forecast horizon"). The only limitations beyond the judicial exception are the two receiving steps, which are insignificant extra-solution data gathering (MPEP 2106.05(g)), and the generic computing device, cloud infrastructure node, processor, and computer-readable medium, which are mere instructions to apply the exception on generic computer components (MPEP 2106.05(f)); considered individually and as an ordered combination, with the claim considered as a whole, these additional elements do not integrate the exception into a practical application and do not amount to significantly more.
Applicant's reliance on the memorandum's reminder that a rejection should be made only when ineligibility is more likely than not (Remarks, page 9) is also unavailing, because eligibility here is not a close call. The improvement analysis set forth in the non-final Office Action and maintained in the final Office Action, including the analysis under Ex parte Desjardins, stands unrebutted by the present Remarks; controlling Federal Circuit authority addresses machine-learning-based prediction claims directly (Recentive Analytics, Inc. v. Fox Corp.); and the specification confirms that the recited forecasting techniques are conventional ([0032], [0045]). Unpatentability under 35 U.S.C. 101 has been established by a preponderance of the evidence. For at least these reasons, the rejection of claims 1-20 under 35 U.S.C. 101 is maintained.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KUANG FU CHEN whose telephone number is (571)272-1393. The examiner can normally be reached M-F 9:00-5:30pm ET.
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, Jennifer Welch can be reached at (571) 272-7212. 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.
/KC CHEN/Primary Patent Examiner, Art Unit 2143
Prosecution Insights