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 .
Notice to Applicant
The following is a Final Office action. In response to Examiner’s Non-Final Rejection of 04/08/2025, Applicant, on 06/27/2025, amended claims 1, 7, 11 to 12, 16 to 17; canceled claims 5 to 6, 13, and 19. Claims 1-4, 7-12, 14-18, and 20 are pending in this application and have been rejected below.
Response to Arguments
Applicant's arguments filed 06/27/2025 have been fully considered, but they are not fully persuasive. The updated 35 USC § 103 and 101 rejections of claims 1-4, 7-12, 14-18, and 20 are applied in light of Applicant's amendments.
Applicant’s arguments with respect to the rejection to the claims of 35 U.S.C. 103 have been considered but are moot because the arguments do not apply to the current combination of references being used in the current rejection. In light of Applicants amendments and arguments the Examiner updated the search and provided new art to reject the claim limitations.
In response, the Examiner respectfully disagrees. The claimed subject matter, is directed to an abstract idea by reciting mathematical relationships, mathematical formulas or equations, mathematical calculations which falls into the “Mathematical concepts” group; and by reciting concepts performed in the human mind (including an observation, evaluation, judgment, opinion), which falls into the “Mental Process” group within the enumerated groupings of abstract ideas set forth in the 2019 PEG.
The claimed subject matter is merely claims a method for calculating and analyzing information regarding networks. Although it may be intended to be performed in a digital environment, the claimed subject matter (as currently claimed in the independent claim) speaks to the calculating and analyzing (modeling and projecting) data. Such steps are not tied to the technological realm, but rather utilizing technology to perform the abstract idea (math). Additionally, the claimed subject matter can also be categorized as a Mental Process as it recites concepts performed in the human mind (observation and evaluation). The steps of calculating data, training/updating models, and generating a model can be performed by a human (mental process/pen and paper). The practice of calculating information and constructing models with set parameters and timelines can be performed without computers, and thus are not tied to technology nor improving technology.
The solution mentioned in the amended limitation is not implemented/integrated into technology and thus not an improvement to the technical field. Further, there is no integration into a practical application as the claims can be interpreted as humans per se, as the claims fail to tie the steps to technology; insignificant extra solution activities (which are merely calculating and/or analyzing data).
The steps relied upon by the Applicant as recited does not improve upon another technology, the functioning of the computer itself, or allow the computer to perform a function not previously performable by a computer. The claims do not mention to any use of a specialized computer and/or processor. The Applicant is using generic computing components (processors) to perform in a generic/expected way (obtaining and analyzing data).The abstract idea is not particular to a technological environment, but is merely being applied to a computer realm. The process of calculating and analyzing data specifically for service project(s), and performing additional analysis can be done without a computer, and thus the claims are not “necessarily rooted", but rather they are utilizing computer technology to perform the abstract idea. The Examiner does not recognize any elements of the Applicant's claims and/or specification that would improve or allow the computer to perform a function(s) not previously performable by the computer, or improve the functioning of the computer itself. It is insufficient to indicate that the claims are novel and non-obvious, and thus contain “something more.” Just because the components may perform a specialized function does not mean that that the computer components are specialized. As such the application of the abstract idea of collecting and analyzing data regarding a network, and performing correlation analysis is insufficient to demonstrate an improvement to the technology.
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 1-4, 7-12, 14-18, and 20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-patentable subject matter. The claims are directed to an abstract idea without significantly more.
Claims 1-4, 7-12, 14-18, and 20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The judicial exception is not integrated into a practical application. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The eligibility analysis in support of these findings is provided below, in accordance with the “2019 Revised Patent Subject Matter Eligibility Guidance” (published on 1/7/2019 in Fed. Register, Vol. 84, No. 4 at pgs. 50-57, hereinafter referred to as the “2019 PEG”).
With respect to Step 1 of the eligibility inquiry (as explained in MPEP 2106), it is first noted that the method (claims 1-10), computer program product (claims 11-15), and system (claims 16-20) are directed to potentially eligible categories of subject matter (i.e., process, machine, and article of manufacture respectively). Thus, Step 1 is satisfied.
With respect to Step 2, and in particular Step 2A Prong One of 2019 PEG, it is next noted that the claims recite an abstract idea by reciting mathematical relationships, mathematical formulas or equations, mathematical calculations which falls into the “Mathematical concepts” group; and by reciting concepts performed in the human mind (including an observation, evaluation, judgment, opinion), which falls into the “Mental Process” group within the enumerated groupings of abstract ideas set forth in the 2019 PEG.
A claim that recites a mathematical calculation, when the claim is given its broadest reasonable interpretation in light of the specification, will be considered as falling within the "mathematical concepts" grouping. A mathematical calculation is a mathematical operation (such as multiplication) or an act of calculating using mathematical methods to determine a variable or number, e.g., performing an arithmetic operation such as exponentiation. There is no particular word or set of words that indicates a claim recites a mathematical calculation. That is, a claim does not have to recite the word "calculating" in order to be considered a mathematical calculation. For example, a step of "determining" a variable or number using mathematical methods or "performing" a mathematical operation may also be considered mathematical calculations when the broadest reasonable interpretation of the claim in light of the specification encompasses a mathematical calculation.
The limitations reciting the abstract idea(s) (mathematical concepts), as set forth in exemplary claim 1, are: training a first sub-model using a subset of a plurality of time series of data values based on a set of periodicities of the time series of data values, each of the time series comprising a series of data values indexed in day order and corresponding to a performance indicator of the network, wherein the first sub-model comprises a type of generalized additive model, and wherein training the type of generalized additive model comprises determining parameters within a plurality of univariate functions of the first sub- model, including a first function indicating a trend of the subset of the time series of data values, a second function indicating seasonality of the subset of the time series of data values, and a third function indicating effects of holidays and events, and wherein the parameters are determined based on reducing modeling error and complexity penalty; training a second sub-model using the subset of the time series of data values captured from the network, wherein the second sub-model comprises a type of autoregressive integrated moving average (ARIMA) model, wherein training the type of ARIMA model comprises normalizing the subset of the time series of data values; determining(mathematical concepts)causing performing remedial measures on the network based on predicting a data value of the performance indicator of the network at a later day, the data value predicted using the hybrid model. (mental process)Independent claims 11 and 16 recite the CRM and system for performing the method of independent claim 1 without adding significantly more. Thus, the same rationale/analysis is applied.
With respect to Step 2A Prong Two of the 2019 PEG, the judicial exception is not integrated into a practical application. The additional elements are directed to A computer implemented method to predict network performance of a network, the method comprising… captured from the network…; An electronic device (as recited in claims 1, 11, and 16). However, these elements fail to integrate the abstract idea into a practical application because they fail to provide an improvement to the functioning of a computer or to any other technology or technical field, fail to apply the exception with a particular machine, fail to apply the judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition, fail to effect a transformation of a particular article to a different state or thing, and fail to apply/use the abstract idea in a meaningful way beyond generally linking the use of the judicial exception to a particular technological environment.
Accordingly, because the Step 2A Prong One and Prong Two analysis resulted in the conclusion that the claims are directed to an abstract idea, additional analysis under Step 2B of the eligibility inquiry must be conducted in order to determine whether any claim element or combination of elements amount to significantly more than the judicial exception.
With respect to Step 2B of the eligibility inquiry, it has been determined that the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The additional limitation(s) is/are directed to: A computer implemented method to predict network performance of a network, the method comprising…; An electronic device (as recited in claims 1, 11, and 16) for implementing the claim steps/functions. These elements have been considered, but merely serve to tie the invention to a particular operating environment (i.e., computer-based implementation), though at a very high level of generality and without imposing meaningful limitation on the scope of the claim.
In addition, Applicant’s Specification (paragraph [0074]) describes generic off-the-shelf computer-based elements for implementing the claimed invention, and which does not amount to significantly more than the abstract idea, which is not enough to transform an abstract idea into eligible subject matter. Such generic, high-level, and nominal involvement of a computer or computer-based elements for carrying out the invention merely serves to tie the abstract idea to a particular technological environment, which is not enough to render the claims patent-eligible, as noted at pg. 74624 of Federal Register/Vol. 79, No. 241, citing Alice, which in turn cites Mayo. See, e.g., Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) (using a telephone for image transmission); OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015) (sending messages over a network); buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014) (computer receives and sends information over a network).
In addition, when taken as an ordered combination, the ordered combination adds nothing that is not already present as when the elements are taken individually. There is no indication that the combination of elements integrate the abstract idea into a practical application. Their collective functions merely provide conventional computer implementation. Therefore, when viewed as a whole, these additional claim elements do not provide meaningful limitations to transform the abstract idea into a practical application of the abstract idea or that the ordered combination amounts to significantly more than the abstract idea itself. Further, the courts have found the presentation of data to be a well-understood, routine, conventional activity, OIP Techs., 788 F.3d at 1362-63, 115 USPQ2d at 1092-93 (see MPEP 2106.05(d)).
The dependent claims (2-10, 12-15, and 17-20) are directed to the same abstract idea as recited in the independent claims, and merely incorporate additional details that narrow the abstract idea via additional details of the abstract idea. For example claims 2-10 “determining (504) the set of periodicities of the time series of data values, wherein the determination comprises performing a Fast Fourier Transform (FFT) on the time series of data values; identifying (602) missing data values in the subset of the plurality of time series of data values; and applying (604) linear interpolation to add in one or more missing data values into the subset of the plurality of time series of data values prior to training the first and second sub-models; :identifying and removing (606) data values that deviate from expected values for a time series of data values over a threshold prior to training the first and second sub- models; wherein training the first sub-model comprises determining parameters for a first function indicating a trend of the subset of the time series of data values, a second function indicating seasonality of the subset of the time series of data values, and a third function indicating effects of holidays and events, and wherein the parameters are determined based on reducing modeling error and complexity penalty; wherein training the second sub-model comprises normalizing the subset of the time series of data values; wherein normalizing the subset of the time series of data values uses a Box-Cox transformation; wherein training the second sub-model comprises using an Akaike information criterion to determine parameter values of the ARIMA model; wherein the second sub-model comprises a Trigonometric Box-Cox transform, ARMA errors, Trend, and Seasonal components (TBATS) model; wherein the performance indicator is one of the following: a call drop rate, a network throughput, a traffic latency, a packet loss rate, a retransmission rate, a reference signal received power (RSRP) level measured by a wireless device in the network, a number of connected wireless devices to a network node, a total number of calls during a period at the network node, and network uptime measured at the network node ”, without additional elements that integrate the abstract idea into a practical application and without additional elements that amount to significantly more to the claims. The remaining dependent claims (12-15 and 17-20) recite the CRM and system for performing the method of claims 2-10. Thus, the same rationale/analysis is applied. Thus, all dependent claims have been fully considered, however, these claims are similarly directed to the abstract idea itself, without integrating it into a practical application and with, at most, a general purpose computer that serves to tie the idea to a particular technological environment, which does not add significantly more to the claims.
The ordered combination of elements in the dependent claims (including the limitations inherited from the parent claim(s)) add nothing that is not already present as when the elements are taken individually. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide conventional computer implementation. Accordingly, the subject matter encompassed by the dependent claims fails to amount to significantly more than the abstract idea itself.
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 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, 4, 7, 10-12, 15-17, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent 11281969 (hereinafter “Rang”) et al., in view of U.S. Patent 9900790 to (hereinafter “Sheen”) et al., in further view of U.S. PGPub 20200379529 to (hereinafter “Le Goff”) et al.
As per claim 1, Rang teaches A computer implemented method to predict… the method comprising:
training a first sub-model using a subset of a plurality of time series of data values based on a set of periodicities of the time series of data values, each of the time series comprising a series of data values indexed in day order and corresponding to a performance indicator …wherein the first sub-model comprises a type of generalized additive model, and wherein training the first sub-model comprises determining parameters within a plurality of univariate functions of the first sub-model; Rang 02: “ Time series data sets are used in a variety of application domains, including for example weather forecasting, finance, econometrics, medicine, control engineering, data center resource management…031-038: the customers or clients on whose behalf the forecasting is being performed may indicate (e.g., using interfaces 177) preferences regarding the metrics to be used to rate the quality of the forecasts. For example, one client may programmatically indicate that the accuracy of the median (50th percentile) forecast for T days in the future is of greatest interest, while for another client, the accuracy of the 90.sup.th percentile forecast for 2*T days of the future may be more important… an additive model may, at least in principle, be used to describe the data. In an additive model, the observed time series is assumed to be the sum of a plurality of independent components around some average or “level,” such as (in the example shown in FIG. 2) a trend component, a seasonal component, and a random or irregular component. In contrast, if the amplitude of the seasonal variations and/or random fluctuations change substantially over time, a multiplicative model may be used (in which the observed values are assumed to be the product of the components rather than the sum), or the time series data may be transformed (e.g., using logarithms) to conform to an additive modeling methodology”
training a second sub-model using the subset of the time series of data values, wherein the second sub-model comprises a type of autoregressive integrated moving average (ARIMA) model; Rang 024: ”FIG. 1 illustrates an example forecasting system in which probabilistic demand forecasts may be generated for time series using composite machine learning models that include a shared neural network sub-model and per-time-series state space sub-models, according to at least some embodiments. As shown, system 100 may include various resources and artifacts of a forecasting service 150 at which a number of different types of forecasting models may be trained and executed for time series data sets. An algorithm and model library 160 of the forecasting service 150 may comprise, for example, composite models 162 with a shared RNN sub-model and per-time-series state space sub-models, other RNN-based models 164 which do not use structural assumptions or state space sub-models, and so on. The state space sub-models may in effect incorporate structural assumptions regarding the time series, such as assumptions regarding trends, seasonality and the like, into the composite forecasting model in the depicted embodiment, and thereby reduce the number of time series time step values that have to be analyzed to make accurate predictions. Some forecasting models that do not include RNNs, such as models 166, may also be implemented in the depicted embodiment at the forecasting service 150. For example, non-RNN models 166 may include regression-only models, exponential smoothing models or autoregressive integrated moving average (ARIMA) models, and so on in some embodiments. The forecasting service may also be referred to as a forecaster in some embodiments. In at least one embodiment, a composite forecasting model comprising a convolutional neural network (CNN) sub-model (e.g., instead of or in addition to an RNN) and one or more per-time-series state space models may be trained and used at the forecasting service.”
wherein training the type of ARIMA model comprises normalizing the subset of the time series of data values; Rang 026: “Feature metadata 120 may include, for example, such information as inventory item product categories, prices, calendar events such as holidays which may affect demands for various items, promotion periods (e.g., time periods in which specific items were on sale for a lower-than-normal price), periods in which an inventory item was out of stock, and so on. With respect to time series pertaining to resource consumption at a data center, e.g., at a cloud computing environment, the feature metadata 120 may include, for example, specific applications for which resource usage data is captured, and so on. Generally speaking, feature metadata 120 may comprise elements of information that could potentially help explain the variation in values over time for the type of time series being considered, and it may therefore be useful to incorporate the feature metadata into the predictive models used for forecasting in at least some embodiments. In some embodiments, the raw metadata may be processed or transformed before it is provided as input to the composite model—e.g., numeric values of the metadata may be normalized, vectorized and so on, categorical values may be transformed to numeric values, and so on…061: According to at least some embodiments, e.g., in order to deal with input time series that deviate from Gaussian distribution assumptions, Box-Cox transformations (or a similar power transformation technique) may be used in a version of a composite forecasting model similar to that introduced above. In one such embodiment, the input time series (observations) may be transformed to more Gaussian-like data via Box-Cox transformation. During training, parameters of the Box-Cox transformation may also be learned by the shared RNN model, jointly with other parameters of the model. Before providing a response to a forecast request, probabilistic prediction values generated by the trained version of the composite model may be transformed back to the domain of the untransformed input data by applying the inverse Box-Cox transformation in such embodiments.”
determining a weight distribution between the first sub-model and second sub- model using additional data values from the time series of data values to generate a hybrid model incorporating the first and second sub-models; Rang 055: “ a real-valued vector of the last layer (e.g., comprising LSTMs) of the RNN sub-model may be mapped to the parameters Θ.sub.i,t of the state space sub-model for the i.sup.th time series by applying affine mappings followed by suitable elementwise transformations constraining the parameters to appropriate ranges. The parameters of the state space sub-models may then be used to compute the likelihood 313 of the given observations z.sub.i,t, which in turn may be used for learning the RNN parameters Φ using the loss function. In at least some embodiments, the state space sub-model parameters may be constrained using the following approach. The output of the RNN sub-model at time step t is denoted as σ.sub.tϵ[AltContent: rect].sup.H. For any state space sub-model parameter θ.sub.t, an affine transformation θ′.sub.t=w.sub.θ.sup.To.sub.t+b.sub.θ may be computed with separate weights w.sub.θϵ[AltContent: rect].sup.H and biases b.sub.θ for each parameter θ. All of these weights and biases may be included in Φ and learned in some embodiments. θ′.sub.t may then be transformed in at least one embodiment to the domain of the parameter by applying, for example, the following transformations: (a) for real-valued parameters such as b.sub.t, no transformation may be required; (b) for positive parameters, the softplus function θ.sub.t=log(1+exp(θ′.sub.t)) may be used; and/or (c) for bounded parameters θ∈[p, q], a scaled and shifted sigmoid θ.sub.t=((q−p)/(1+exp(−θ′.sub.t)))+p may be employed. In practice, in some embodiments stricter bounds than those theoretically required may be imposed; for example, imposing an upper bound on the observation noise variance or a lower bound on the innovation strengths may help to stabilize the training of the composite model in the presence of outliers.”
and predicting a data value of the performance indicator …at a later day using the hybrid model;Rang 057: “FIG. 4 illustrates example aspects of a prediction phase of a forecasting model which includes a shared recurrent neural network sub-model and per-time-series state space sub-models, according to at least some embodiments. Once the parameters Φ of the RNN sub-model have been learned, they may be used to generate probabilistic forecasts for individual time series. In at least some embodiments, prediction samples may be generated from the state space sub-model corresponding to a given time series. First, for example, the posterior of the latent state p(l.sub.T|z.sub.1:T) for the last time step Tin the training range (to the left of the forecast start boundary 417 in FIG. 4) for a given time series may be computed in such embodiments, and then the state transition formulation and the observation model (the probabilistic value generation formulation) may be applied recursively to generate the prediction samples. The inputs 405 to the shared RNN sub-model for the training range may comprise the covariate features in the depicted embodiment as discussed earlier in the context of FIG. 3, and the output 407 of the shared RNN sub-model may be used to compute the state space sub-model parameters 409, with the time series observations 411 being used in loss function computations as before.”
Rang may not explicitly teach the following. However, Sheen teaches:
captured from the network based…network performance of a network…of the network…; Sheen, Abstract: “A system and method of predicting cellular network performance based on observed performance indicator data from cells in the cellular network. The method includes accessing a set of observed performance indicator data, the performance indicator data including a time sequenced measure of performance indicators for the cellular network. The method then classifies the observed performance data based on a cell as one of a high load growth cell and a high load non-growth cell. Based on the classification of the cell, the method computes a future value of at least one performance indicator using a predictive model based on testing data for the cell in the observed performance indicator data. The predictive model is derived from training data in the observed performance indicator data. An indication of the future value of the at least one of the performance indicators is output when the future value exceeds an alarm value.”
Rang and Sheen are deemed to be analogous references as they are reasonably pertinent to each other and directed towards measuring, collecting, and analyzing information with a series of inputs to solve similar problems in the similar environments. Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to have modified Rang with the aforementioned teachings from Sheen with a reasonable expectation of success, by adding steps that allow the software to utilize network data with the motivation to more efficiently and accurately organize and analyze data [Sheen, Abstract].
Rang and Sheen may not explicitly teach the following. However, Le Goff teaches:
and wherein training the type of generalized additive model comprises determining parameters within a plurality of univariate functions of the first sub- model, including a first function indicating a trend of the subset of the time series of data values, a second function indicating seasonality of the subset of the time series of data values, and a third function indicating effects of holidays and events, and wherein the parameters are determined based on reducing modeling error and complexity penalty;Le Goff 0099-0120: “ the prediction model 335 is constructed at fixed intervals, for example every day, every few days, every few weeks, and the like. The raw forecast 340 is then updated, from time to time or on a continuous basis, by calculating a moving average based on past estimation errors, which are differences between past estimates obtained using the raw forecast 340 and past real measurements that may be stored in the local cache 352. Instead or in addition, it is contemplated that models such as seasonal autoregressive integrated moving average (SARIMA) or Holt-Winters may be used to extract fitted autoregressive, moving average and integration parameters and time-polynomial trend to yield a state-space representation of the prediction model. These parameters may then be applied to latest received measurements to obtain a dynamic forecast. The prediction model 335 is gradually updated to follow a trend of the measurements over time. In the same or another embodiment, the machine learning system may be retrained using recent measurements when it is found that the prediction model 335, although updated based on latest measurements, consistently provides estimates that fail to predict the actual measurements...0120: the machine learning system 525 may detect and ignore outliers in the accumulated measurements, for example measurements that are outside of an i.sup.th percentile of measurements accumulated over a period of one hour, a value i for the percentile being a number less than 100, for example 70 percent.”
Rang, Sheen, and Le Goff are deemed to be analogous references as they are reasonably pertinent to each other and directed towards measuring, collecting, and analyzing information with a series of inputs to solve similar problems in the similar environments. Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to have modified Rang and Sheen with the aforementioned teachings from Le Goff with a reasonable expectation of success, by adding steps that allow the software to utilize seasonal data with the motivation to more efficiently and accurately organize and analyze data [Le Goff, 0120 ].
As per claim 2, Rang, Sheen, and Le Goff teach all the limitations of claim 1.
Rang and Sheen may not explicitly teach the following. However, Le Goff teaches:
determining (504) the set of periodicities of the time series of data values, wherein the determination comprises performing a Fast Fourier Transform (FFT) on the time series of data values;Le Goff 0184: “The method (600) of any one of clauses 1 to 15, wherein the machine learning system (525) constructs the prediction model by applying, on the stored measurements, a forecasting algorithm selected from an autoregressive integrated moving average (ARIMA), a triple exponential smoothing (Holt-Winters), a Fast Fourier transform (FFT) decomposition, a current state redefinition, a polynomial combination, a linear regression, a multilayer perceptron (MLP), a long short-term memory (LSTM), a Gaussian distribution, and a combination thereof.”
Rang, Sheen, and Le Goff are deemed to be analogous references as they are reasonably pertinent to each other and directed towards measuring, collecting, and analyzing information with a series of inputs to solve similar problems in the similar environments. Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to have modified Rang and Sheen with the aforementioned teachings from Le Goff with a reasonable expectation of success, by adding steps that allow the software to utilize FFT with the motivation to more efficiently and accurately organize and analyze data [Le Goff, 0015 ].
As per claim 4, Rang, Sheen, and Le Goff teach all the limitations of claim 1.
Rang and Sheen may not explicitly teach the following. However, Le Goff teaches:
identifying and removing data values that deviate from expected values for a time series of data values over a threshold prior to training the first and second sub- models;Le Goff 0093-0120: “These scripts are periodically executed in the metric platform 130 to create metric aggregation data series that are stored in the live metric storage 132 and/or in the persistent metric storage 134 may be provided to an analysis function 140 that executes a continuous monitoring 142 of the measurements. The analysis and monitoring may lead to the generation of alerts 144 and/or to the detection of anomalies 146 in view of initiating corrective actions. The alerts 144 may be presented in text or graphical form… These events could potentially affect the accuracy of the prediction model for one or more impacted servers 230. In response to these events, the second data processing function 355 may issue a monitoring switch signal 358 to prevent an action of the anomaly detection function 365 for the one or more impacted servers 230. Otherwise stated, the second data processing function 355 may detect that the prediction model can no longer be relied on and cause the anomaly detection function 365 to revert to other anomaly detection mechanisms, for example by comparing the latest measurements 314 to fixed thresholds, detecting and anomaly when a threshold is exceeded for a predetermined time period such as a few hours, or detecting an anomaly when a threshold is exceeded for at least a number of servers 230… the machine learning system 525 may detect and ignore outliers in the accumulated measurements, for example measurements that are outside of an i.sup.th percentile of measurements accumulated over a period of one hour, a value i for the percentile being a number less than 100, for example 70 percent.”
Rang, Sheen, and Le Goff are deemed to be analogous references as they are reasonably pertinent to each other and directed towards measuring, collecting, and analyzing information with a series of inputs to solve similar problems in the similar environments. Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to have modified Rang and Sheen with the aforementioned teachings from Le Goff with a reasonable expectation of success, by adding steps that allow the software to utilize threshold data with the motivation to more efficiently and accurately organize and analyze data [Le Goff, 0120 ].
As per claim 7, Rang, Sheen, and Le Goff teach all the limitations of claim 1.
In addition, Rang teaches:
wherein normalizing the subset of the time series of data values uses a Box-Cox transformation; Rang 026: “Feature metadata 120 may include, for example, such information as inventory item product categories, prices, calendar events such as holidays which may affect demands for various items, promotion periods (e.g., time periods in which specific items were on sale for a lower-than-normal price), periods in which an inventory item was out of stock, and so on. With respect to time series pertaining to resource consumption at a data center, e.g., at a cloud computing environment, the feature metadata 120 may include, for example, specific applications for which resource usage data is captured, and so on. Generally speaking, feature metadata 120 may comprise elements of information that could potentially help explain the variation in values over time for the type of time series being considered, and it may therefore be useful to incorporate the feature metadata into the predictive models used for forecasting in at least some embodiments. In some embodiments, the raw metadata may be processed or transformed before it is provided as input to the composite model—e.g., numeric values of the metadata may be normalized, vectorized and so on, categorical values may be transformed to numeric values, and so on…061: According to at least some embodiments, e.g., in order to deal with input time series that deviate from Gaussian distribution assumptions, Box-Cox transformations (or a similar power transformation technique) may be used in a version of a composite forecasting model similar to that introduced above. In one such embodiment, the input time series (observations) may be transformed to more Gaussian-like data via Box-Cox transformation. During training, parameters of the Box-Cox transformation may also be learned by the shared RNN model, jointly with other parameters of the model. Before providing a response to a forecast request, probabilistic prediction values generated by the trained version of the composite model may be transformed back to the domain of the untransformed input data by applying the inverse Box-Cox transformation in such embodiments.”
As per claim 10, Rang, Sheen, and Le Goff teach all the limitations of claim 1.
In addition, Sheen teaches:
wherein the performance indicator is one of the following: a call drop rate, a network throughput, a traffic latency, a packet loss rate, a retransmission rate, a reference signal received power (RSRP) level measured by a wireless device in the network, a number of connected wireless devices to a network node, a total number of calls during a period at the network node, and network uptime measured at the network node; Sheen 013-018: “The network 100 may comprise any wired or wireless network that provides communication connectivity for devices. The network 100 may include various cellular network and packet data network components such as a base transceiver station (BTS), a node-B, an evolved node-B (eNodeB), a base station controller (BSC), a radio network controller (RNC), a service GPRS support node (SGSN), a gateway GPRS support node (GGSN), a WAP gateway, mobile switching center (MSC), short message service centers (SMSC), a home location registers (HLR), a visitor location registers (VLR), an Internet protocol multimedia subsystem (IMS), and/or the like… in a mobile data service network, the service accessibility may be determined through the Packet Data Protocol (PDP) Context Activation Success Rate KPI, which may be an aggregated ratio of the successful PDP context activations to PDP context attempts…PS.Service.Downlink.Average.Throughput—The average downlink service traffic throughput at cell level;… The total downlink traffic volume for packet data convergence protocol (PDCP) service data units (SDUs) in a cell.”
Rang and Sheen are deemed to be analogous references as they are reasonably pertinent to each other and directed towards measuring, collecting, and analyzing information with a series of inputs to solve similar problems in the similar environments. Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to have modified Rang with the aforementioned teachings from Sheen with a reasonable expectation of success, by adding steps that allow the software to utilize network data with the motivation to more efficiently and accurately organize and analyze data [Sheen, Abstract].
Claims 11 and 16 are directed to the device and CRM for performing the method of claim 1 above. Since Rang, Sheen, and Le Goff each the device and CRM, the same art and rationale apply.
Claims 15 and 20 are directed to the device and CRM for performing the method of claim 10 above. Since Rang, Sheen, and Le Goff each the device and CRM, the same art and rationale apply.
Claims 12 and 17 are directed to the device and CRM for performing the method of claims 2 above. Since Rang, Sheen, and Le Goff teach the device and CRM, the same art and rationale apply.
Claims 3 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent 11281969 (hereinafter “Rang”) et al., in view of U.S. Patent 9900790 to (hereinafter “Sheen”) et al., in further view of U.S. PGPub 20200379529 to (hereinafter “Le Goff”) et al., and in further view of U.S. Patent 11775936 to (hereinafter “Zhang”) et al.
As per claim 3, Rang, Sheen, and Le Goff teach all the limitations of claim 1.
Rang and Sheen may not explicitly teach the following. However, Zhang teaches:
identifying missing data values in the subset of the plurality of time series of data values; and applying linear interpolation to add in one or more missing data values into the subset of the plurality of time series of data values prior to training the first and second sub-models;Zhang, Abstract and 0044-0058: “Techniques for forecasting long duration floating holidays in online traffic are described. According to some embodiments, a machine learning service receives a request to train a time series forecast model on time series data of a user, receives an input for the time series forecast model that comprises a first feature weight that represents a first pivot day and a second feature weight that represents a second pivot day, performs a linear interpolation on the first feature weight and the second feature weight for a day between the first pivot day and the second pivot day to generate a linearly interpolated first weight of the first feature weight for a feature vector and a linearly interpolated second weight of the second feature weight for the feature vector, determines a first coefficient for the time series forecast model based at least in part on the time series data of the user, the linearly interpolated first weight of the first feature weight from the feature vector, and the linearly interpolated second weight of the second feature weight from the feature vector, generates, by the time series forecast model comprising the first coefficient, a prediction for a future day, and transmits the prediction to the user..”
Rang, Sheen, Le Goff and Zhang are deemed to be analogous references as they are reasonably pertinent to each other and directed towards measuring, collecting, and analyzing information with a series of inputs to solve similar problems in the similar environments. Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to have modified Rang, Sheen, and Le Goff with the aforementioned teachings from Zhang with a reasonable expectation of success, by adding steps that allow the software to interpolate with the motivation to more efficiently and accurately organize and analyze data [Zhang 0044].
Claim 18 is directed to the CRM for performing the method of claim 3 above. Since Rang, Sheen, Le Goff, and Zhang teach the CRM, the same art and rationale apply.
Claims 8 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent 11281969 (hereinafter “Rang”) et al., in view of U.S. Patent 9900790 to (hereinafter “Sheen”) et al., in further view of U.S. PGPub 20200379529 to (hereinafter “Le Goff”) et al., and in further view of U.S. PGPub 20180302291 to (hereinafter “Srin”) et al.
As per claim 8, Rang, Sheen, and Le Goff teach all the limitations of claim 1.
Rang, Sheen, and Le Goff may not explicitly teach the following. However, Srin teaches:
wherein training the second sub-model comprises using an Akaike information criterion to determine parameter values of the ARIMA model; Srin, Table 2: “Box-cox transformations may help in balancing the seasonal fluctuations and the random variation across the data/time series. Auto ARIMA The auto.ARIMA (Auto ARIMA) function in R in uses a variation of the Hyndman and Khandakar algorithm which combines principles such as minimization of AIC (Akaike information criteria) and unit root tests to obtain the best ARIMA model. ARIMA(p, d, q) models may be selected from the following: ARIMA(2, d, 2), ARIMA(0, d, 0), ARIMA(1, d, 0), ARIMA(0, d, 1). Note: 1) The number of differences “d” is determined using repeated KPSS tests 2) The values of free parameters p and q are then chosen by minimizing the AICs after differencing the data “d” times.”
Rang, Sheen, Le Goff, and Srin are deemed to be analogous references as they are reasonably pertinent to each other and directed towards measuring, collecting, and analyzing information with a series of inputs to solve similar problems in the similar environments. Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to have modified Rang, Sheen, and Le Goff with the aforementioned teachings from Srin with a reasonable expectation of success, by adding steps that allow the software to utilize AIC with the motivation to more efficiently and accurately organize and analyze data [Srin 0052].
Claim 14 is directed to the CRM for performing the method of claim 8 above. Since Rang, Sheen, Le Goff, and Srin teach the CRM, the same art and rationale apply.
Claims 9 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent 11281969 (hereinafter “Rang”) et al., in view of U.S. Patent 9900790 to (hereinafter “Sheen”) et al., in further view of U.S. PGPub 20200379529 to (hereinafter “Le Goff”) et al., and in further view of U.S. PGPub 20180300737 to (hereinafter “Bledsoe”) et al.
As per claim 9, Rang, Sheen, and Le Goff teach all the limitations of claim 1.
Rang, Sheen, and Le Goff may not explicitly teach the following. However, Bledsoe teaches:
wherein training the second sub-model comprises using an Akaike information criterion to determine parameter values of the ARIMA model; Bledsoe, Table 3: “Number of Samples/Observations per Forecasting Model Forecasting Model Samples (N) ARIMA + Covariate Regressor (Xreg) 1 ARIMA + Xreg + Weekly 1 Autoregressive Integrated Moving Average (ARIMA) 1 ARIMA + Weekly 1 ETS 1 Error, Trend, Seasonal (ETS) + Weekly 1 Box-Cox transform, ARMA errors, Trend, and 1 Seasonal components (BATS) + Weekly Trigonometric, Box-Cox transform, ARMA errors, 1 Trend, and Seasonal components (TBATS) + Weekly TBATS + Annual 1 TBATS + Weekly + Annual 1 Auto-regressive Neural Network 3 Auto-regressive Neural Network + Xreg 3 Bayesian Structural Time Series 3 Bayesian Structural Time Series .”
Rang, Sheen, Le Goff, and Bledsoe are deemed to be analogous references as they are reasonably pertinent to each other and directed towards measuring, collecting, and analyzing information with a series of inputs to solve similar problems in the similar environments. Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to have modified Rang, Sheen, and Le Goff with the aforementioned teachings from Bledsoe with a reasonable expectation of success, by adding steps that allow the software to utilize mathematical concepts with the motivation to more efficiently and accurately organize and analyze data [Bledsoe 0107].
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
LIU; Changxin. DECISION-MAKING METHOD OF COMPREHENSIVE ALUMINA PRODUCTION INDEXES BASED ON MULTI-SCALE DEEP CONVOLUTIONAL NETWORK, .U.S. PGPub 20210192272 The invention relates to the technical field of decision-making for comprehensive alumina production, in particular to a decision-making method of comprehensive alumina production indexes based on a multi-scale deep convolutional network.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period wil