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 .
Status of Claims
This action is in response to the amendments filed 11/18/2025. Claims 1-9 and 13-14 have been amended, claims 10-12 have been cancelled. Claims 1-9 and 13-14 are currently pending.
Response to Arguments
Claims 10-12 have been cancelled, therefore the rejections of claims 10-12 no longer stand.
In light of Applicant’s amendment, the objection to claims 1 and 13-14 has been withdrawn.
Applicant’s arguments regarding the 101 rejection have been fully considered but they are not persuasive. Applicant argues that the amended limitations related to a “learning model generation system” and acquiring time series data as learning data “in a case where it is determined that the probability distribution has converged”, performing learning of a learning model to produce a trained model, and wherein “the learning data of the acquired time series data corresponds to an amount of data. . .so that a possibility of the trained model requiring relearning is reduced” integrate any claimed judicial exceptions into a practical application “corresponding to an improvement to the technical field of training data models as applied to factory automation systems”. Examiner respectfully disagrees and notes that the claims do not recite any particular “learning model” or particular steps of training a learning model that would reflect this alleged improvement, and that the claims do not recite any particular application to factory automation systems. Examiner further notes that the limitation “wherein the learning data of the acquired time series data corresponds to an amount of data. . .so that a possibility of the trained model requiring relearning is reduced” has been interpreted as the intended outcome of “performing learning of a learning model. . .” and does not provide any additional patentable weight to this claim (see MPEP 2111.04). The 101 rejections have been updated to include the amended limitations and to clarify the reasoning given for the limitations that were not amended.
Applicant’s arguments regarding the prior art rejection have been fully considered but they are not persuasive. Applicant argues that the Mori reference does not teach acquiring learning data “in a case where it is determined that the probability distribution has converged” and cites page 4573 section B paragraph 2 to argue that “the convergence of the SRs (stopping rules) is not required for the framework’s operation”. Examiner respectfully disagrees and notes that this paragraph of Mori merely discusses different options for different kinds of databases, and that section 2145(X)(D)(1) of the MPEP states “the prior art’s mere disclosure of more than one alternative does not constitute a teaching away from any of these alternatives because such disclosure does not criticize, discredit, or otherwise discourage the solution claimed….”.
Applicant further argues that the Mori reference does not explicitly teach “wherein the learning data of the acquired time series data corresponds to an amount of data. . .so that a possibility of the trained model requiring relearning is reduced”. Examiner respectfully disagrees and notes that this limitation has been interpreted as the intended outcome of “performing learning of a learning model. . .” and does not provide any additional patentable weight to this claim (see MPEP 2111.04). The prior art rejections have been updated to include the amended limitations and to clarify the reasoning given for the limitations that were not amended.
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 1-9 and 13-14 are rejected under 35 U.S.C. 101. Claims 1-9 are directed to a system, claim 13 is directed to an additional method, and claim 14 is directed to and additional non-transitory computer readable medium; therefore, claims 1-14 fall within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter). However, claims 1-14 fall within the judicial exception of an abstract idea, specifically the abstract ideas of “Mental Processes” (including observation, evaluation, and opinion) and “Mathematical Concepts (including mathematical calculations and relationships)”.
Claim 1:
Claim 1 is directed to a system; therefore, the claim does fall within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
Claim 1 recites the following abstract ideas:
divide the time series data into a plurality of pieces of substring data; generate a plurality of substring data sets that are sets of the substring data (mental step directed to observation, evaluation – a person could divide observed time series data and generate a plurality of substring datasets from the divided time series data in their mind, potentially assisted by pen and paper (see MPEP 2106.04(a)(2)(III));
calculate a feature amount of the substring data (mental step directed to evaluation – a person could calculate a feature amount of the substring data in their mind);
generate a probability distribution of the feature amount for each of the substring data set; and determine whether or not the probability distribution has converged (mental step directed to observation, evaluation – a person could generate a probability distribution of a feature amount for each substring data set and determine whether an observed or determined probability distribution has converged in their mind);
perform learning of a learning model using the learning data and generates a trained model, wherein the learning data of the acquired time series data corresponds to an amount of data for learning which generates the trained model, so that a possibility of the trained model required relearning is reduced (as the claim does not require the “learning model” to be any particular kind of artificial intelligence or machine learning model, nor do the claims recite particular technical steps for generating or training this model, performing learning of a model using learning data and generating a trained model are interpreted as mental steps – a person could generate and learn (or train) a model in their mind based on observed learning data. Examiner notes that the limitation “wherein the learning data of the acquired time series data corresponds to an amount of data. . .so that a possibility of the trained model requiring relearning is reduced” has been interpreted as the intended outcome of “performing learning of a learning model. . .” and does not provide any additional patentable weight to this claim (see MPEP 2111.04)).
Claim 1 recites the following additional elements:
processing circuitry configured to acquire time series data, and acquire the time series data as learning data in a case where it is determined that the probability distribution has converged. The processing circuitry is interpreted as generic computer components used to merely implement the claimed abstract ideas, and acquiring the time series data as learning data in a case wherein it is determined that a probability has converged is interpreted as an additional element directed to receiving data over a network under a given condition. These additional elements do not integrate the abstract ideas into a practical application or amount to significantly more than the abstract ideas (see MPEP 2106.05(d)).
Claim 13 is an additional method claim and its limitation is included in claim 1. The only difference is that claim 13 requires a method. Therefore, claim 13 is rejected for the same reasons as claim 1.
Claim 14 is an additional non-transitory computer readable medium claim and its limitation is included in claim 1. The only difference is that claim 14 requires a non-transitory computer readable medium. Therefore, claim 14 is rejected for the same reasons as claim 1.
The independent claims are not patent eligible.
Dependent claims 2-9 when analyzed as a whole are held to be patent ineligible under 35 U.S.C. 101 because the additional recited limitations fail to establish that the claims are not directed to an abstract idea, as they recite further embellishment of the judicial exception.
Claims 2-5 recite generating substring data sets by adding substring data which is either overlapping or not overlapping with other substring datasets. These limitations are all interpreted as mental steps – a person could generate a first, second, and third substring data set in their mind and decide whether the substring data sets should include common data or not in their mind.
Claim 6 recites wherein the processing circuitry generates a first group having a plurality of the substring data sets and a second group having a same number of the substring data sets as the first group and having at least one substring data set not included in the first group, the processing circuitry calculates a similarity between the probability distribution of the substring data set included in the first group and the probability distribution of the substring data set included in the second group, and the processing circuitry determines that the probability distribution has converged in a case where the similarity has converged.
Generating a first and second group of substring datasets, wherein the second group contains at least one substring not included in the first group, calculating a similarity between probability distributions of the groups, and determining that the similarity has converged are all interpreted as mental steps – a person could generate groups of substring datasets, calculate a similarity between observed or determined probability distributions of the groups, and determine that the similarity has converged in their mind. The processing circuity is interpreted as a generic computer component used to implement the claimed abstract ideas and does not integrate the claimed abstract ideas or amount to significantly more than the claimed abstract ideas (see MPEP 2106.05(f)).
Claim 7 recites wherein the processing circuitry calculates the feature amount for each of the substring data. Calculating a feature amount for each of the substring data is interpreted as a mental step – a person could calculate a feature amount for each substring data in their mind. The processing circuity is interpreted as a generic computer component used to implement the claimed abstract ideas and does not integrate the claimed abstract ideas or amount to significantly more than the claimed abstract ideas (see MPEP 2106.05(f)).
Claim 8 recites wherein the processing circuitry calculates a comparison value between the first substring data and the second substring data as the feature amount. Calculating a comparison value between first and second substring data as the feature amount is interpreted as a mental step – a person could calculate a comparison value between first and second substring data as a feature amount in their mind. The processing circuity is interpreted as a generic computer component used to implement the claimed abstract ideas and does not integrate the claimed abstract ideas or amount to significantly more than the claimed abstract ideas (see MPEP 2106.05(f)).
Claim 9 recites wherein the processing circuitry generates a first set including the plurality of the substring data sets from the time series data included from a first time to a second time, and generates a second set including the plurality of the substring data sets from the time series data included from a third time to a fourth time, and the processing circuitry determines that an amount of the time series data is sufficient in a case where a predetermined condition is met in both the first set and the second set. Generating a first substring data set based on time series data from a first to a second time, generating a second substring data set based on time series data from a third to a fourth time, and determining that an amount of time series data is sufficient when a predetermined condition is met for both sets are all interpreted as mental steps – a person could generate a first and second substring data set based on time series data from a first to second and third to fourth time and determine whether the amount of time series data is sufficient based on determining whether a predetermined condition is met for both sets in their mind. The processing circuity is interpreted as a generic computer component used to implement the claimed abstract ideas and does not integrate the claimed abstract ideas or amount to significantly more than the claimed abstract ideas (see MPEP 2106.05(f)).
Viewed as a whole, these additional claim elements do not provide meaningful limitations to transform the abstract idea into a patent eligible application of the abstract idea such that the claims amount to significantly more than the abstract idea itself. Therefore, the claims are rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1, 3, 5, 7, and 13-14 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Mori et al* (“Early Classification of Time Series by Simultaneously Optimizing the Accuracy and Earliness”, herein Mori).
*a copy of this document was provided with the IDS dated 10/25/2023.
Regarding claim 1, Mori teaches a learning model generation system (section III A para. 2 recites “we will train a set of classifier {ht}Lt = 1 for all timestamps t ϵ {1, 2, . . ., L} or for a user defined subset of timestamps” (i.e., a system which generates a learning model)) comprising: processing circuitry configured to acquire time series data; divide the time series data into a plurality of pieces of substring data; generate a plurality of substring data sets that are sets of the substring data (section II para. 10 recites “Fig. 1-2 and section III A para. 1 recite “we will use a training set X = {(TS1, CL1), (TS2, CL2), . . . , (TSn, CLn)} of labeled full-length time series of finite length” (i.e., acquired time series data that has been divided to generate a plurality of substring, or subsets of data));
calculate a feature amount of the substring data; generate a probability distribution of the feature amount for each of the substring data sets (section III A para. 1 recites “we will use a training set X = {(TS1, CL1), (TS2, CL2), . . . , (TSn, CLn)} of labeled full-length time series of finite length” (i.e., the label CL1 corresponds to the data subset TS1). Section III A para. 3 recites “To train these classifiers, we will use the whole training set X. Each classifier ht will receive the first t points of a series, and will output the posterior probabilities for each class at that time. These classifiers will be used to obtain the posterior class membership probabilities for the new unlabeled (test) time series at each time t. We can see an illustrative example of the construction of the ht classifiers in Fig. 1” (i.e., determining a feature, or a label, for each data subset and generating the probabilities of each feature, or label, for each data subset));
determine whether or not the probability distribution has converged (section III A para. 6-7 recite “we analyze a basic SR (SR1γ ) (i.e., a stopping rule), based on intuition and defined by the following linear rule: {EQ2} where pt = (pt1, pt2, . . . , ptk) is the vector of posterior probabilities for the k possible classes issued by the corresponding ht for a given time series, pt1:k and pt2:k are the first and second largest posterior probability values obtained at time t, and γ = (γ1, γ2, γ3) is a vector of parameters that takes real values between −1 and 1. The interpretation of this SR is the following: if the rule outputs a value of 1, we conclude that the prediction is reliable enough, and thus, the class corresponding to the maximum posterior probability value is provided. On the contrary, a value of 0 indicates that the prediction is not yet reliable, and so, we should wait until a larger part of the time series is available. If the entire time series is available and the SR has not triggered, the class prediction obtained at t = L is used” (i.e., determining that convergence condition, or stopping rule is met for the each of the time-series subsets t is met));
acquire the time series data as learning data in a case where it is determined that the probability distribution has converged (section III A para. 6-7 recite “we analyze a basic SR (SR1γ ) (i.e., a stopping rule), based on intuition and defined by the following linear rule: {EQ2} where pt = (pt1, pt2, . . . , ptk) is the vector of posterior probabilities for the k possible classes issued by the corresponding ht for a given time series, pt1:k and pt2:k are the first and second largest posterior probability values obtained at time t, and γ = (γ1, γ2, γ3) is a vector of parameters that takes real values between −1 and 1. The interpretation of this SR is the following: if the rule outputs a value of 1, we conclude that the prediction is reliable enough, and thus, the class corresponding to the maximum posterior probability value is provided. On the contrary, a value of 0 indicates that the prediction is not yet reliable, and so, we should wait until a larger part of the time series is available. If the entire time series is available and the SR has not triggered, the class prediction obtained at t = L is used” (i.e., determining that convergence condition, or stopping rule, for acquired time series data is met));
and perform learning of a learning model using the learning data and generates a trained model, wherein the learning data of the acquired time series data corresponds to an amount of data for learning which generates the trained model, so that a possibility of the trained model required relearning is reduced (section III A para. 2 recites “we will train a set of classifier {ht}Lt = 1 for all timestamps t ϵ {1, 2, . . ., L} or for a user defined subset of timestamps” (i.e., generating a trained learning model and using it on the acquired time series data. Examiner notes that the limitation “wherein the learning data of the acquired time series data corresponds to an amount of data. . .so that a possibility of the trained model requiring relearning is reduced” has been interpreted as the intended outcome of “performing learning of a learning model. . .” and does not provide any additional patentable weight to this claim (see MPEP 2111.04))).
Regarding claim 3, Mori teaches the learning model generation system according to claim 1, wherein the processing circuitry generates a first substring data set and a second substring data set not including the substring data common to the first substring data set (fig. 1-2 and section III A para. 1 recite “we will use a training set X = {(TS1, CL1), (TS2, CL2), . . . , (TSn, CLn)} of labeled full-length time series of finite length” (i.e., first and second subsets of data are distinct from one another)).
Regarding claim 5, Mori teaches the learning model generation system according to claim 3, wherein the processing circuitry generates a first substring data set and a third substring data set not including substring data common to the second substring data set (Mori fig. 1-2 and section III A para. 1 recite “we will use a training set X = {(TS1, CL1), (TS2, CL2), . . . , (TSn, CLn)} of labeled full-length time series of finite length” (i.e., first, second, and at least a third subset of data are distinct from one another)).
Regarding claim 7, Mori teaches the learning model generation system according to claim 1, wherein the processing circuitry calculates the feature amount for each of the substring data (Mori section III A para. 1 recites “we will use a training set X = {(TS1, CL1), (TS2, CL2), . . . , (TSn, CLn)} of labeled full-length time series of finite length”. Mori section III A para. 3 recites “To train these classifiers, we will use the whole training set X. Each classifier ht will receive the first t points of a series, and will output the posterior probabilities for each class at that time. These classifiers will be used to obtain the posterior class membership probabilities for the new unlabeled (test) time series at each time t. We can see an illustrative example of the construction of the ht classifiers in Fig. 1” (i.e., determining a feature, or a label, for each data subset)).
Regarding claim 9, Mori teaches the learning model generation system according to claim 1, wherein the processing circuitry generates a first set including the plurality of the substring data sets from the time series data included from a first time to a second time, and generates a second set including the plurality of the substring data sets from the time series data included from a third time to a fourth time (Mori fig. 1-2 and section III A para. 1 recite “we will use a training set X = {(TS1, CL1), (TS2, CL2), . . . , (TSn, CLn)} of labeled full-length time series of finite length” (i.e., a first subset TS1 from a first to a second time and another subset of data TSn from at least third to a fourth time – Examiner notes that one of ordinary skill in the art would normally interpret TS2 as being from a second to a third time)),
and the processing circuitry determines that an amount of the time series data is sufficient in a case where a predetermined condition is met in both the first set and the second set (Mori section III A para. 6-7 recite “we analyze a basic SR (SR1γ ) (i.e., a stopping rule), based on intuition and defined by the following linear rule: {EQ2} where pt = (pt1, pt2, . . . , ptk) is the vector of posterior probabilities for the k possible classes issued by the corresponding ht for a given time series, pt1:k and pt2:k are the first and second largest posterior probability values obtained at time t, and γ = (γ1, γ2, γ3) is a vector of parameters that takes real values between −1 and 1. The interpretation of this SR is the following: if the rule outputs a value of 1, we conclude that the prediction is reliable enough, and thus, the class corresponding to the maximum posterior probability value is provided. On the contrary, a value of 0 indicates that the prediction is not yet reliable, and so, we should wait until a larger part of the time series is available. If the entire time series is available and the SR has not triggered, the class prediction obtained at t = L is used” (i.e., determining that an amount of time series data is sufficient, or not all of the data in time series t=length L, when a predetermined condition, or stopping rule for the each of the time-series subsets t is met)).
Claim 13 is a method claim and its limitation is included in claim 1. The only difference is that claim 13 requires a method. Therefore, claim 13 is rejected for the same reasons as claim 1.
Claim 14 is a non-transitory computer readable medium claim and its limitation is included in claim 1. The only difference is that claim 14 requires a non-transitory computer readable medium. Therefore, claim 14 is rejected for the same reasons as claim 1.
Claim Rejections - 35 USC § 103
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 2 and 4 are rejected under 35 U.S.C. 103 as being unpatentable over Mori et al* (“Early Classification of Time Series by Simultaneously Optimizing the Accuracy and Earliness”, herein Mori) in view of Heumann et al (US 20060074828 A1, herein Heumann).
Regarding claim 2, Mori teaches the learning model generation system according to claim 1.
However, Mori does not explicitly teach wherein the processing circuitry generates a second substring data set by adding substring data not including a first substring data set to the first substring data set.
Heumann teaches wherein the processing circuitry generates a second substring data set by adding substring data not including a first substring data set to the first substring data set (Heumann para. [0061] recites “the time-sorted labeled training data D.sub.SORTED is partitioned into four mutually exclusive subsets D.sub.1, D.sub.2, D.sub.3, and D.sub.4 of approximately equal size (i.e., no member of any subset belongs to any other subset)”. Heumann para. [0062] recites “the number M of subsets may vary according to the particular application, and the subsets may also be constructed to overlap such that one or more subsets includes one or more data samples from a subset immediately previous to or immediately subsequent to the given subset in time” (i.e., a given subset of data can include at least some of the data from a previous subset)).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine these teachings by modifying the time-series subsets from Mori with the subsets that may overlap from Heumann. Mori and Heumann are both directed to time-series data analysis methods. As section III A para. 2 of Mori states, “the user could choose any other subset of timestamps, based on domain knowledge or other information of the shape of the series or even by using specific time series sampling methods”, one of ordinary skill in the art find it obvious that a user could choose to use the overlapping subsets from Heumann in place of or in addition to the distinct subsets from Mori.
Regarding claim 4, Mori teaches the learning model generation system according to claim 3.
However, Mori does not explicitly teach wherein the processing circuitry generates the first substring data set and a third substring data set including at least one substring data included in the second substring data set.
Heumann teaches wherein the processing circuitry generates the first substring data set and a third substring data set including at least one substring data included in the second substring data set (Heumann para. [0061] recites “the time-sorted labeled training data D.sub.SORTED is partitioned into four mutually exclusive subsets D.sub.1, D.sub.2, D.sub.3, and D.sub.4 of approximately equal size (i.e., no member of any subset belongs to any other subset)”. Heumann para. [0062] recites “the number M of subsets may vary according to the particular application, and the subsets may also be constructed to overlap such that one or more subsets includes one or more data samples from a subset immediately previous to or immediately subsequent to the given subset in time” (i.e., a given subset of data can include at least some of the data from a previous subset without other subsets necessarily having to share data)).
See claim 2 for motivation to combine.
Claims 6 and 8 are rejected under 35 U.S.C. 103 as being unpatentable over Mori et al* (“Early Classification of Time Series by Simultaneously Optimizing the Accuracy and Earliness”, herein Mori) in view of McCallum (US 20200074982 A1, herein McCallum).
Regarding claim 6, Mori teaches the learning model generation system according to claim 1, wherein the processing circuitry generates a first group having a plurality of the substring data sets and a second group having a same number of the substring data sets as the first group and having at least one substring data set not included in the first group (Mori fig. 1-2 and section III A para. 1 recite “we will use a training set X = {(TS1, CL1), (TS2, CL2), . . . , (TSn, CLn)} of labeled full-length time series of finite length” (i.e., first and second subsets of data are distinct from one another and equally sized)), and whether a probability distribution has converged (section III A para. 6-7 recite “we analyze a basic SR (SR1γ ) (i.e., a stopping rule), based on intuition and defined by the following linear rule: {EQ2} where pt = (pt1, pt2, . . . , ptk) is the vector of posterior probabilities for the k possible classes issued by the corresponding ht for a given time series, pt1:k and pt2:k are the first and second largest posterior probability values obtained at time t, and γ = (γ1, γ2, γ3) is a vector of parameters that takes real values between −1 and 1. The interpretation of this SR is the following: if the rule outputs a value of 1, we conclude that the prediction is reliable enough, and thus, the class corresponding to the maximum posterior probability value is provided. On the contrary, a value of 0 indicates that the prediction is not yet reliable, and so, we should wait until a larger part of the time series is available. If the entire time series is available and the SR has not triggered, the class prediction obtained at t = L is used” (i.e., determining that convergence condition, or stopping rule, for a probability distribution of time series data is met)).
However, Mori does not teach wherein the processing circuitry calculates a similarity between the probability distribution of the substring data set included in the first group and the probability distribution of the substring data set included in the second group, and a calculated similarity.
McCallum teaches wherein the processing circuitry calculates a similarity between the probability distribution of the substring data set included in the first group and the probability distribution of the substring data set included in the second group, and a calculated similarity (McCallum para. [0018] recites “Any number and/or type(s) of method(s), algorithm(s), circuit(s), etc. may be used to detect beats in the incoming digital audio 106. For example, an example approach includes the probabilistic tracking of regularly occurring peaks in a spectral flux signal” (i.e., probabilistic features of the data can be determined, potentially using a probabilistic classifier like the one from Mori). McCallum para. [0022] recites “12. The neural network 104 has an example deep feature generator 122 that generates, develops, forms, computes, etc. so called deep features 124 that can be combined e.g., by a distance calculator 126 of some sort, that generates a distance metric that can be used to embed and/or classify audio, data, objects, information, etc. The deep features 124 computed by the deep feature generator 122 may represent classes and/or descriptors of audio, data, objects, information, etc. When deep features 124 for different portions of the incoming digital audio 106 are compared by the distance calculator 126, the distance metric 126 can determine whether the portions are musically similar or musically dissimilar”. McCallum para. [0023] recites “The neural network 104 updates the internal coefficients 128 so the deep features 124 generated from the similar data and the anchor data become closer together (e.g., in Euclidean distance), and deep features 124 generated from the dissimilar data and the anchor data become further apart (e.g., in Euclidean distance)” (i.e., calculating a similarity between data subsets)).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine these teachings by modifying the time-series classification method from Mori with the subset similarity comparison from McCallum to provide additional feature analysis of the time-series subsets from Mori. Mori and McCallum are both directed to time-series data analysis methods, and paragraph [018] of McCallum states “Any number and/or type(s) of method(s), algorithm(s), circuit(s), etc. may be used to detect beats in the incoming digital audio 106. For example, an example approach includes the probabilistic tracking of regularly occurring peaks in a spectral flux signal” (i.e., audio data being understood as a form of time-series data) and that probabilistic . As Mori does not require any particular form of input time-series data, and the similarity comparisons from McCallum can be used in probabilistic analysis like the probabilistic classifiers from Mori, one of ordinary skill in the art would understand how to combine these methods such that the classification method from Mori could provide more detailed feature analysis of its time-series data subsets.
Regarding claim 8, Mori teaches the learning model generation system according to claim 1.
However, Mori does not explicitly teach calculating a comparison value between the first substring data and the second substring data as a feature amount
McCallum teaches calculating a comparison value between the first substring data and the second substring data as a feature amount (McCallum para. [0022] recites “12. The neural network 104 has an example deep feature generator 122 that generates, develops, forms, computes, etc. so called deep features 124 that can be combined e.g., by a distance calculator 126 of some sort, that generates a distance metric that can be used to embed and/or classify audio, data, objects, information, etc. The deep features 124 computed by the deep feature generator 122 may represent classes and/or descriptors of audio, data, objects, information, etc. When deep features 124 for different portions of the incoming digital audio 106 are compared by the distance calculator 126, the distance metric 126 can determine whether the portions are musically similar or musically dissimilar” (i.e., determining comparison values between different subsets of data to determine features of the data)).
See claim 6 for motivation to combine.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 20150112900 A1 (Ariyoshi et al) teaches a method for acquiring observation values that continue at predetermined time intervals, as prediction data, from time-series data, generating a prediction model to calculate values from the time-series data.
US 20200397346 A1 (Nakajima) teaches a method for acquiring time-series sensor data and segmenting the data into partially overlapping subsets.
US 11562297 B2 (Goldszmidt et al) teaches a method for triggering an update or retraining of a machine-learning model upon detecting that a distribution of particular input data set is sufficiently different from a distribution training input data set used to previously train the model.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LEAH M FEITL whose telephone number is (571) 272-8350. The examiner can normally be reached on M-F 0900-1700 EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Viker Lamardo can be reached on (571) 270-5871. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/L.M.F./ Examiner, Art Unit 2147
/VIKER A LAMARDO/Supervisory Patent Examiner, Art Unit 2147