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 the Claims
Claims 1-4, 10-13, 16-19, and 21 have been amended. Claims 1-21 are currently pending and have been considered by the Examiner.
Claim Objections
Claims 1 and 7-8 are objected to because of the following informalities:
In claim 1, lines 12-15 should be nested under line 11.
In claim 7, line 2, “Guassian” appears to be a misspelling of “Gaussian”. In claim 8, line 2, Examiner suggests capitalizing the term “gaussian” as in claim 7. Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 2-4, 11-13, and 17-19 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claim 2 recites the limitations “compare the compressed input to a reconstructed input generated from the compressed input; and predict the directionality based on the comparison” in lines 6-8. These limitations lack support in the written disclosure. Specification paragraphs [0038] and [0042]-[0043] disclose comparing the normalized input data to the reconstructed input generated from the compressed input. The comparison is made during a training step. [0069]-[0070] discloses predicting the directionality based on the training step comprising the comparison. Normalized input data is equivalent to “normalized data values” as recited in the claims.
Claims 3-4 are rejected for failing to cure the deficiencies of claim 2.
Claim 11 recites the same new matter as claim 2 and is rejected for at least the same reasons. Claims 12-13 are rejected for failing to cure the deficiencies of claim 11.
Claim 17 recites the same new matter as claim 2 and is rejected for at least the same reasons. Claims 18-19 are rejected for failing to cure the deficiencies of claim 17.
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 2-4, 11-13, and 17-19 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
In claim 2, the limitation in lines 6-7 renders the claim indefinite because it is unclear whether this limitation is supposed to recite “compare the normalized data values to the reconstructed input generated from the compressed input”. Examiner treats claims 6-7 as if they had recited the above interpretation.
Claims 3-4 are rejected for failing to cure the deficiencies of claim 2.
Claim 11 recites the same indefinite limitation as claim 2 and is rejected for at least the same reasons. Claims 12-13 are rejected for failing to cure the deficiencies of claim 11.
Claim 17 recites the same indefinite limitation as claim 2 and is rejected for at least the same reasons. Claims 18-19 are rejected for failing to cure the deficiencies of claim 17.
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 1-23 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1: Claims 1-9 recite a system comprising a processor (a system), claims 10-15 recite a method, claims 16-20 recite a non-transitory computer readable medium (a product), and claims 21-23 recite a system comprising a processor (a system). Each of a system, a method, and a product fall under one of the four statutory categories of patent eligible subject matter.
Claim 1
Step 2A Prong 1: Decompose the time series of data into a plurality of clusters to generate the mixture model, each cluster from among the plurality of clusters comprising a normal distribution of a respective subset of the plurality of data values from the time series of data is a judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
For each data value in the time series of data: identify a corresponding cluster, from among the plurality of clusters of the mixture model, against which the data value is to be normalized is an observation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Determine a normalization value for the corresponding cluster is a mathematical calculation. In specification paragraph [0035], Equation 3 discloses a formula for normalizing a data value. Determining a normalization value includes calculating a mean or variance of the closest cluster.
Normalize the data value based on the normalization value is a mathematical calculation. In specification paragraph [0035], Equation 3 discloses a formula for normalizing a data value.
Generate,
Step 2A Prong 2: A mixture model that approximates a non-normal distribution of sequential data amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f).
A machine-learning model trained on one or more sets of time series of data amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f).
A processor programmed to: access a time series of data having a plurality of data values that exhibit a non-normal distribution, each data value from among the plurality of data values corresponding to a point in time in the time series of data amounts to an insignificant extra-solution activity under MPEP 2106.05(g). A processor is a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f).
Provide the normalized data values to the machine-learning model trained to predict a directionality of the time series of data amounts to invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f).
Using the machine-learning model amounts to invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f).
The additional elements as disclosed above, alone or in combination, do not integrate the abstract ideas into a practical application as they are mere insignificant extra solution activities as disclosed in combination with generic computer functions that are implemented to perform the abstract ideas disclosed above. The claim is directed to an abstract idea.
Step 2B: A mixture model that approximates a non-normal distribution of sequential data amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f).
A machine-learning model trained on one or more sets of time series of data amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f).
A processor programmed to: access a time series of data having a plurality of data values that exhibit a non-normal distribution, each data value from among the plurality of data values corresponding to a point in time in the time series of data is analogous to retrieving information from memory, which the courts have recognized as a well-understood, routine, conventional activity under MPEP 2106.05(d)(II). A processor is a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f).
Provide the normalized data values to the machine-learning model trained to predict a directionality of the time series of data amounts to invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f).
Using the machine-learning model amounts to invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f).
The additional elements as disclosed above, in combination with the abstract ideas, are not sufficient to amount to significantly more than the abstract ideas as they are well-understood, routine and conventional activities as disclosed in combination with generic computer functions that are implemented to perform the abstract ideas disclosed above. The claim is not patent eligible.
Claim 2 incorporates the rejection from claim 1.
Step 2A Prong 1: The abstract ideas from claim 1 are incorporated. Generate the prediction is a judgement mental and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Encode,
Compare the compressed input to a reconstructed input generated from the compressed input is an observation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Predict the directionality based on the comparison is a judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Step 2A Prong 2 and Step 2B: The machine-learning model comprises an autoencoder amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f).
The processor amounts to a generic computer component for applying the abstract ideas on a computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim 3 incorporates the rejection from claim 2.
Step 2A Prong 1: The abstract ideas from claim 2 are incorporated. Generate the compressed input based on the normalized data values is a mathematical calculation and a judgement mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Step 2A Prong 2 and Step 2B: The autoencoder comprises a trained encoder amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim 4 incorporates the rejection from claim 3.
Step 2A Prong 1: The abstract ideas from claim 3 are incorporated. Generate the reconstructed input from the compressed input generated by the encoder is a mathematical calculation and a judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Step 2A Prong 2 and Step 2B: The autoencoder comprises a trained decoder amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim 5 incorporates the rejection from claim 1.
Step 2A Prong 1: The abstract ideas from claim 1 are incorporated. Identify the corresponding normal distribution is an observation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Determine a distance between the data value to each normal distribution from among the plurality of normal distributions is a mathematical calculation. In specification paragraph [0035], the sentence in lines 5-6 discloses the distance may be a difference between the particular data value and the mean of the k-cluster.
Select the corresponding normal distribution that is closest to the data value based on the determined distances is a judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Step 2A Prong 2 and Step 2B: The processor amounts to a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim 6 incorporates the rejection from claim 1.
Step 2A Prong 1: The abstract ideas from claim 1 are incorporated. Identify a number of the plurality of clusters to be used is an observation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
The time series of data is approximated based on the plurality of clusters is a judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Step 2A Prong 2 and Step 2B: The processor amounts to a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim 7 incorporates the rejection from claim 1.
Step 2A Prong 1: The abstract ideas from claim 1 are incorporated. Decompose the time series of data into a Gaussian mixture comprising overlapping clusters of normal distributions is a judgement and evaluation mental process based on a mathematical calculation which can reasonably be performed in the human mind with the aid of pencil and paper.
Step 2A Prong 2 and Step 2B: The processor amounts to a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim 8 incorporates the rejection from claim 1.
Step 2A Prong 1: The abstract ideas from claim 1 are incorporated. Decompose the time series of data into a gaussian mixture comprising non-overlapping clusters of normal distributions is a judgement and evaluation mental process based on a mathematical calculation which can reasonably be performed in the human mind with the aid of pencil and paper.
Step 2A Prong 2 and Step 2B: The processor amounts to a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim 9 incorporates the rejection from claim 1.
Step 2A Prong 1: The abstract ideas from claim 1 are incorporated. Determine the normalization value by determine a mean or variance of the corresponding cluster is a mathematical calculation. In specification paragraph [0035], Equation 3 discloses a normalization value is a mean or variance of the closest cluster.
Step 2A Prong 2 and Step 2B: The processor amounts to a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claims 10-15 each recites a method which incorporates the same features as the system of claims 1-6, respectively, and are therefore rejected for at least the same reasons.
Claim 16 recites a product which incorporates the same features as the system of claim 1 and is therefore rejected for at least the same reasons.
In Step 2A Prong 2 and Step 2B, a non-transitory computer-readable storage medium storing instructions that, when executed by a processor, programs the processor amount to generic computer components for applying the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claims 17-20 each recites a product which incorporates the same features as the system of claims 2-5, respectively, and are therefore rejected for at least the same reasons.
Claim 21
Step 2A Prong 1: Merge the output from each of the plurality of RNNs is a judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
Generate a prediction based on the merged output is a judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper. The claim recites an abstract idea.
Step 2A Prong 2 and Step 2B: A plurality of recursive neural networks (RNNs) configured to operate in parallel to collectively form a parallel neural network architecture, each neural network from among the plurality of RNNs comprising: an input layer that receives a time series of data, one or more RNN layers, and one or more dense layers amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f).
A processor programmed to: provide each RNN, from among the plurality of RNNs, with a respective time series of data, each respective time series of data comprising sequential data values that vary independently of one another over time amounts to invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f).
Obtain an output from a last one of the one or more dense layers of each RNN amounts to invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f).
The additional elements as disclosed above, alone or in combination, do not integrate the abstract ideas into a practical application as they are mere generic computer functions that are implemented to perform the abstract ideas disclosed above. The claim is directed to an abstract idea.
The additional elements as disclosed above, in combination with the abstract ideas, are not sufficient to amount to significantly more than the abstract ideas as they are generic computer functions that are implemented to perform the abstract ideas disclosed above. The claim is not patent eligible.
Claim 22 incorporates the rejection from claim 21.
Step 2A Prong 1: The abstract ideas from claim 21 are incorporated. Generate a mixture model for each of the respective time series of data, the mixture model comprising a plurality of clusters of normal distributions that together approximates the respective time series of data is a judgement and evaluation mental process based on mathematical calculations which can reasonably be performed in the human mind with the aid of pencil and paper.
Step 2A Prong 2 and Step 2B: The processor amounts to a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim 23 incorporates the rejection from claim 22.
Step 2A Prong 1: The abstract ideas from claim 22 are incorporated. Normalize values of each of the respective time series of data based on the mixture model generated for the respective time series of data is a mathematical calculation. In specification paragraph [0035], Equation 3 discloses a formula for normalizing a data value.
Step 2A Prong 2 and Step 2B: The processor amounts to a generic computer component for applying the abstract ideas on a generic computer under MPEP 2106.05(f). The claim is not patent eligible.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 5-7, 9-10, 14-16, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Pallath et al. (US 20180150547 A1) in view of Abbaszadeh et al. (US 20200067969 A1) and Ouyang et al. (US 20190036795 A1). All references were cited in the PTO-892 issued 02/12/2026.
Regarding claim 1, Pallath teaches: A system, comprising: a
a machine-learning model trained on one or more sets of time series of data; and ([0062]-[0069] discloses a trained Random Forest algorithm)
a processor programmed to: ([0104], lines 1-2)
access a time series of data having a plurality of data values that exhibit a non-normal distribution, each data value from among the plurality of data values corresponding to a point in time in the time series of data; ([0023], lines 1-5; [0027] on page 3, col. 1, lines 1-12; [0040] discloses receiving time series data that exhibit non-normal underlying structures.)
decompose the time series of data into a plurality of clusters to generate the
for each data value in the time series of data: identify a corresponding cluster, … ([0048], lines 1-3)
provide the
generate, using the machine-learning model, a prediction relating to the directionality and/or a magnitude of the time series of data. ([0037], final 2 lines and [0097]. A “directionality” is an increase or decrease in the usage relative to the previous time points, and a “magnitude” is a value of the usage.)
However, Pallath does not explicitly teach: generate the mixture model;
identify a corresponding cluster, from among the plurality of clusters of the mixture model, against which the data value is to be normalized, determine a normalization value for the corresponding cluster, and normalize the data value based on the normalization value;
provide the normalized data values
But Abbaszadeh teaches: generate the mixture model, each cluster from among the plurality of clusters comprising a normal distribution ([0096] and [0097], lines 1-10)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have used Abbaszadeh’s Gaussian mixture model (GMM) for Pallath’s clustering. A motivation for the combination is that a GMM may allow for the building of a complex probability distribution from a linear superposition of simpler components. Gaussian distributions may be the most common choice as mixture components because of the mathematical simplicity of parameter estimation as well as their ability to perform well in many situations. (Abbaszadeh, [0096])
However, Pallath and Abbaszadeh do not explicitly teach: identify a corresponding cluster, from among the plurality of clusters of the mixture model, against which the data value is to be normalized, determine a normalization value for the corresponding cluster, and normalize the data value based on the normalization value;
provide the normalized data values
But Ouyang teaches: identify a corresponding cluster, from among the plurality of clusters
provide the normalized data values ([0083], lines 1-2 and reference claim 6, lines 8-9)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have standardized every data point in a particular cluster in Pallath and Abbaszadeh, and to have applied the standardized data points. A motivation for the combination is that standardization makes the data points within a cluster more comparable to one another.
Regarding claim 5, the combination of Pallath, Abbaszadeh, and Ouyang teaches: The system of claim 1,
Pallath teaches: wherein to identify the corresponding normal distribution, the processor is programmed to: determine a distance between the data value to each normal distribution from among the plurality of normal distributions; and ([0048], lines 3-14; [0051], lines 1-6)
select the corresponding normal distribution that is closest to the data value based on the determined distances. ([0030] and [0051], lines 1-6)
Regarding claim 6, the combination of Pallath, Abbaszadeh, and Ouyang teaches: The system of claim 1,
Pallath teaches: wherein the processor is further programmed to: identify a number of the plurality of clusters to be used, wherein the time series of data is approximated based on the plurality of clusters. (All of [0047] and [0048], lines 1-6)
Regarding claim 7, the combination of Pallath, Abbaszadeh, and Ouyang teaches: The system of claim 1,
Pallath teaches: wherein to decompose the time series of data, the processor is further programmed to decompose the time series of data into
However, Pallath does not explicitly teach: a Gaussian mixture comprising overlapping clusters of normal distributions.
But Abbaszadeh teaches: a Gaussian mixture comprising overlapping clusters of normal distributions. ([0098], lines 1-3)
A motivation for the combination is the same as the motivation given for claim 1.
Regarding claim 9, the combination of Pallath, Abbaszadeh, and Ouyang teaches: The system of claim 1, wherein, to determine the normalization value, the processor is programmed to:
Pallath teaches: determine a mean or variance of the corresponding cluster. ([0029], lines 18-end and [0030] discloses determining a centroid of a corresponding cluster, which is a mean value.)
Abbaszadeh at [0097], lines 4-10 discloses that a centroid is a mean. Determining the normalization value is recited as an intended effect of determining a mean. In the combination of references, Pallath’s system determines a centroid, and then Ouyang’s system determines the normalization value.
Claims 10 and 14-15 each recites a method which incorporates the same features as the system of claims 1 and 5-6, respectively, and are therefore rejected for at least the same reasons.
Claim 16 recites a product which incorporates the same features as the system of claim 1 and is therefore rejected for at least the same reasons.
Pallath teaches: A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, programs the processor to: ([0104], lines 4-8)
Claim 20 recites a product which incorporates the same features as the system of claim 5 and is therefore rejected for at least the same reasons.
Claims 2-4, 11-13, and 17-19 are rejected under 35 U.S.C. 103 as being unpatentable over Pallath et al. (US 20180150547 A1, cited in the PTO-892 issued 02/12/2026) in view of Abbaszadeh et al. (US 20200067969 A1, cited in the PTO-892 issued 02/12/2026), Ouyang et al. (US 20190036795 A1, cited in the PTO-892 issued 02/12/2026), and Mizutani et al. (US 20210263954 A1).
Regarding claim 2, the combination of Pallath, Abbaszadeh, and Ouyang teaches: The system of claim 1,
Pallath teaches: … and wherein to generate, using the machine-learning model, the prediction, the processor is further programmed to: encode, input based on the
predict the directionality
However, Pallath and Abbaszadeh do not explicitly teach: wherein the machine-learning model comprises an autoencoder,
encode, by the autoencoder, an outcome based on the normalized data values;
compare the compressed input to a reconstructed input generated from the compressed input;
predict the directionality based on the comparison.
But Ouyang teaches: normalized data values; ([0080]-[0083] and reference claim 6)
A motivation for the combination is the same as the motivation given for claim 1.
However, Pallath, Abbaszadeh, and Ouyang do not explicitly teach: wherein the machine-learning model comprises an autoencoder,
encode, by the autoencoder, an outcome based on the normalized data values;
compare the compressed input to a reconstructed input generated from the compressed input;
predict the directionality based on the comparison.
But Mizutani teaches: wherein the machine-learning model comprises an autoencoder, ([0035])
encode, by the autoencoder, a compressed input based on the [input]
compare the
predict the [label]
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have incorporated Mizutani’s autoencoder and classifier architecture into the combination of Pallath, Abbaszadeh, and Ouyang, where Mizutani’s classifier is analogous to Pallath’s Random Forest algorithm. In the combination of references, Pallath’s time series would be normalized by Ouyang’s technique before being input to Mizutani’s autoencoder. A motivation for the combination is that compressing data before classifying it can reduce consumption of computing resources. (Mizutani, [0002]-[0003])
Regarding claim 3, the combination of Pallath, Abbaszadeh, Ouyang, and Mizutani teaches: The system of claim 2,
Pallath teaches: an encoder
However, Pallath and Abbaszadeh do not explicitly teach: wherein the autoencoder comprises an encoder trained to generate the compressed input based on the normalized data values.
But Ouyang teaches: normalized data values. ([0080]-[0083] and reference claim 6)
A motivation for the combination is the same as the motivation given for claim 1.
However, Pallath, Abbaszadeh, and Ouyang do not explicitly teach: wherein the autoencoder comprises an encoder trained to generate the compressed input based on the normalized data values.
But Mizutani teaches: wherein the autoencoder comprises an encoder trained to generate the compressed input based on the
A motivation for the combination is the same as the motivation given for claim 2.
Regarding claim 4, the combination of Pallath, Abbaszadeh, Ouyang, and Mizutani teaches: The system of claim 3,
However, Pallath, Abbaszadeh, and Ouyang do not explicitly teach: wherein the autoencoder comprises a decoder trained to generate the reconstructed input from the compressed input generated by the encoder.
But Mizutani teaches: wherein the autoencoder comprises a decoder trained to generate the reconstructed input from the compressed input generated by the encoder. ([0040], [0042], lines 1-3 and [0043], lines 1-7 teaches training an autoencoder comprising a decoder)
A motivation for the combination is the same as the motivation given for claim 2.
Claims 11-13 each recites a method which incorporates the same features as the system of claims 2-4, respectively, and are therefore rejected for at least the same reasons.
Claims 17-19 each recites a product which incorporates the same features as the system of claims 2-4, respectively, and are therefore rejected for at least the same reasons.
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Pallath et al. (US 20180150547 A1) in view of Abbaszadeh et al. (US 20200067969 A1), Ouyang et al. (US 20190036795 A1), and Parlak et al. (US 20200159925 A1). All references were cited in the PTO-892 issued 02/12/2026.
Regarding claim 8, the combination of Pallath, Abbaszadeh, and Ouyang teaches: The system of claim 1,
Pallath teaches: wherein to decompose the time series of data, the processor is further programmed to decompose the time series of data into overlapping clusters of normal distributions. ([0029], lines 12-13 and [0048] discloses applying a K-means clustering algorithm, which is a non-overlapping clustering algorithm.)
Abbaszadeh at [0098], lines 1-3 teaches “GMM is a soft clustering method (i.e., overlapping clusters)”. Therefore, Pallath and Abbaszadeh do not explicitly teach: a gaussian mixture comprising non-overlapping clusters of normal distributions.
But Parlak teaches: a gaussian mixture comprising non-overlapping clusters of normal distributions. ([0032], lines 1-10)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have used Parlak’s GMM to perform hard clustering. A motivation for the combination is to assign a data point to exactly one cluster. (Parlak, [0032])
Claim 21 is rejected under 35 U.S.C. 103 as being unpatentable over Ranjan et al. (US 20220103444 A1) in view of Shamir et al. (US 20210158156 A1). Both references were cited in the PTO-892 issued 02/12/2026.
Regarding claim 21, Ranjan teaches: each neural network
a processor programmed to: provide each RNN, from among the plurality of RNNs, with a respective time series of data, each respective time series of data comprising sequential data values that vary independently of one another over time; ([0047] and [0121]-[0127] discloses providing time-series network model 700 with multivariate time-series data 702. The data values are voltage, temperature, etc. over time. The model 700 is an RNN because it includes recurrent layers. The data values of voltage and temperature vary independently from one another at least because they are different types of measurements.)
obtain an output from a last one of the one or more dense layers of each RNN; ([0127], final 3 lines)
…
generate a prediction based on the
However, Ranjan does not explicitly teach: A system, comprising: a plurality of recursive neural networks (RNNs) configured to operate in parallel to collectively form a parallel neural network architecture, each neural network from among the plurality of RNNs comprising: layers
…
merge the output from each of the plurality of RNNs;
generate a prediction based on the merged output.
But Shamir teaches: A system, comprising: a plurality of recursive neural networks (RNNs) configured to operate in parallel to collectively form a parallel neural network architecture, each neural network from among the plurality of RNNs comprising [layers] ([0043], lines 4-8, [0050], lines 1-5, [0067], lines 1-5, and [0088], lines 3-12 discloses three RNNs 22a-c configured to operate in parallel, each RNN comprising layers.)
…
merge the output from each of the plurality of RNNs; ([0050], lines 1-5 discloses aggregating the respective outputs to generate the ensemble output.)
generate a prediction based on the merged output. ([0050], lines 1-5 discloses aggregating the respective outputs to generate the ensemble output, which is a prediction.)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have generated an ensemble of Ranjan’s RNNs based on Shamir’s techniques. A motivation for the combination is that use of an ensemble improves the reproducibility of the models, making predictions of two independently trained ensembles diverge less from one another. (Shamir, [0038])
Claims 22-23 are rejected under 35 U.S.C. 103 as being unpatentable over Ranjan et al. (US 20220103444 A1) in view of Shamir et al. (US 20210158156 A1) and Abbaszadeh et al. (US 20200067969 A1). All references were cited in the PTO-892 issued 02/12/2026.
Regarding claim 22, the combination of Ranjan and Shamir teaches: The system of claim 21, wherein the processor is further programmed to:
However, Ranjan and Shamir do not explicitly teach: generate a mixture model for each of the respective time series of data, the mixture model comprising a plurality of clusters of normal distributions that together approximates the respective time series of data.
But Abbaszadeh teaches: generate a mixture model for each of the respective time series of data, the mixture model comprising a plurality of clusters of normal distributions that together approximates the respective time series of data. ([0070], lines 6-10, [0096] and [0097], lines 1-10)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have clustered Ranjan and Shamir’s time series data based on Abbaszadeh’s GMM. A motivation for the combination is that a GMM may allow for the building of a complex probability distribution from a linear superposition of simpler components. Gaussian distributions may be the most common choice as mixture components because of the mathematical simplicity of parameter estimation as well as their ability to perform well in many situations. (Abbaszadeh, [0096])
Regarding claim 23, the combination of Ranjan, Shamir, and Abbaszadeh teaches: The system of claim 22, wherein the processor is further programmed to:
Ranjan teaches: normalize values of each of the respective time series of data ([0093], lines 7-14)
However, Ranjan and Shamir do not explicitly teach: normalize values of each of the respective time series of data based on the mixture model generated for the respective time series of data.
Abbaszadeh’s teaches a mixture model generated for each respective time series of data at [0096] and [0097], lines 1-10. It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have applied Ranjan’s standardization to the time series data belonging to each of Abbaszadeh’s clusters. A motivation for the combination is the same as the motivation given for claim 22.
Response to Arguments
The following is Examiner’s response to Applicant’s arguments filed on 05/14/2026.
Applicant’s First Arguments Under 35 U.S.C. 101: On pages 10-11, Applicant argues that under Step 2A, Prong 1, the claims are not directed to a mere mental process, but rather to a specific, computer-implemented data processing pipeline for handling non-normal time-series data. Under Step 2A Prong 2, even if the claims recite a mental process, the claims recite an integration of any alleged abstract idea into a practical application that improves the performance of machine-learning models, particularly for non-normal time-series data.
The claims recite a structured mixture-model-based normalization pipeline that fundamentally alters how input data is processed prior to machine-learning prediction.
By aligning each data value with a corresponding distributional cluster prior to model input, the system produces normalized data that better reflects the underlying structure of non-normal timeseries data, thereby enabling improved prediction of directionality and magnitude. As in Ex parte Desjardins (precedential), the claims are directed to a particularized improvement in machine-learning processing, not to mathematical concepts in the abstract, and therefore integrate any alleged abstract idea into a practical application.
Examiner’s Response: Applicant's arguments have been fully considered but they are not persuasive. The last paragraph on page 10 lists three steps enumerated (i), (ii), and (iii) which allegedly provides a technical improvement. Each of these steps constitutes an abstract idea. Decomposing the time series into clusters representing respective normal distribution is a judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper. Identifying a corresponding cluster for each data value is an observation mental process which can reasonably be performed in the human mind with the aid of pencil and paper. Normalizing each data value using parameters (e.g., mean and variance) of the selected cluster before providing the normalized data to a machine-learning model is a mathematical calculation. In specification paragraph [0035], Equation 3 discloses a formula for normalizing a data value using parameters. Additionally, generating a prediction relating to the directionality and/or magnitude of the time series of data is a judgement mental and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper.
MPEP 2106.05(a) states, “It is important to note, the judicial exception alone cannot provide the improvement.” MPEP 2106.05(a), subsection II states, “it is important to keep in mind that an improvement in the abstract idea itself (e.g. a recited fundamental economic concept) is not an improvement in technology.”
In Step 2A Prong 2, providing the normalized data values to the machine-learning model trained to predict a directionality of the time series of data, and using the machine-learning model each amounts to invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f).
The abstract ideas which transform data for input to the machine-learning model and which generate a prediction are not improvements to machine-learning processing itself. Examiner respectfully disagrees that pending claim 1 is similar to Ex Parte Desjardins. The claims in Desjardins solve a technical problem of catastrophic forgetting in machine learning. The limitations in pending claim 1 as a whole are NOT an analogous factual setting to the claims at issue in Desjardins. Pending claim 1 does not integrate the abstract ideas into a practical application for the reasons provided above.
Applicant’s Second Arguments Under 35 U.S.C. 101: On pages 12-13, Applicant argues claim 21 recites a particular arrangement of multiple RNNs operating in parallel with a specific dataflow and aggregation mechanism, which imposes meaningful limits on how the alleged abstract idea is implemented. The specification makes clear that this architecture addresses technical problems in machine learning systems, including inefficiencies and performance limitations of conventional serial RNN architectures. Specifically, conventional approaches suffer from computational delays and inefficiencies due to sequential processing, whereas the claimed architecture executes multiple RNNs in parallel and merges their outputs to improve performance and predictive capability. Thus, claim 21 recites a technical improvement to computer functionality, namely improving how neural networks process sequential data by using a specific and coordinated parallel architecture with output fusion.
For at least these reasons, the specification sets forth the improvement and the claims reflect this improvement, much like the specification and claims in Desjardin, which was recently deemed precedential. As such, the claims recite eligible subject matter, ending the subject matter eligibility inquiry.
Examiner’s Response: Applicant's arguments have been fully considered but they are not persuasive. In Step 2A Prong 1, the features of merging the output from each of the plurality of RNNs and generating a prediction based on the merged output are both judgement and evaluation mental process which can reasonably be performed in the human mind with the aid of pencil and paper. Merging outputs and generating a prediction alone cannot provide the technical improvements because these features are abstract ideas. See MPEP 2106.05(a).
Claim 21 recites architectural details including a plurality of RNNs each comprising an input layer for receiving input data, RNN layers, and dense layers for generating output data; and a parallel arrangement of the RNNs. These details are generic features of RNNs which amount to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f). They do not integrate the abstract ideas into a practical application. The claim lacks details which show how the parallel architecture would improve efficiency, model performance, and predictive capability, and how the parallel architecture would reduce computational delays when compared to sequential models. The claim lacks details about the functionality and processing aspects of individual layers that would solve the technical problems asserted in the remarks. The claim lacks details about the dataflow (inputs/outputs) through each layer of an individual RNN.
The additional elements as disclosed above, alone or in combination, do not integrate the abstract ideas into a practical application as they are mere insignificant extra solution activities as disclosed in combination with generic computer functions that are implemented to perform the abstract ideas disclosed above. The claim is directed to an abstract idea.
Examiner respectfully disagrees that pending claim 21 is similar to Ex Parte Desjardins for the same reasons set forth in the Examiner’s response to the Applicant’s first arguments.
Applicant’s Third Arguments Under 35 U.S.C. 101: On page 12, Applicant argues that under Step 2B, even if the claims are directed to a mental process under Step 2A (Prongs 1 and 2), they recite a combination of features that provide an inventive concept significantly more than just a mathematical operations or a mental process. In particular, claims 1, 10, and 16 recite a specific sequence of operations including decomposing a time series into clusters representing respective normal distributions, performing per-data-value cluster assignment, determining a normalization value tied to the selected cluster, normalizing each data value based on that cluster-specific normalization, and providing the normalized data values to a machine-learning model to generate a prediction. This ordered combination implements a cluster-conditioned normalization pipeline that transforms input data based on localized statistical structure prior to machine-learning processing, which differs from conventional global normalization approaches. The Office Action does not identify any evidence that such a pipeline, particularly the per-data-value adaptive normalization using a mixture model prior to model input, was well-understood, routine, or conventional.
Examiner’s Response: Applicant's arguments have been fully considered but they are not persuasive. As explained in the Examiner’s response to the Applicant’s first argument, the limitations of decomposing a time series into clusters representing respective normal distributions, performing per-data-value cluster assignment, determining a normalization value tied to the selected cluster, normalizing each data value based on that cluster-specific normalization, and generating a prediction are abstract ideas, which alone cannot provide the technical improvement. In Step 2B, providing the normalized data values to a machine-learning model amounts to invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f). Claim analysis under 101 as described in MPEP 2106.04 does not require the examiner to identify whether abstract ideas are well-understood, routine, and conventional. When an additional element has been identified as an insignificant extra-solution activity in Step 2A Prong 2, the analysis proceeds with evaluating whether the additional element is well-understood, routine, and conventional activity in Step 2B. The additional element “providing the normalized data values to a machine-learning model” was not evaluated as being well-understood, routine, and conventional activity because the Office Action identified it as invoking computers merely as a tool to perform an existing process under MPEP 2106.05(f).
Applicant’s Fourth Arguments Under 35 U.S.C. 101: On page 13, Applicant argues that even if the claim 21 recites an abstract idea under Step 2A (Prongs 1 and 2), the ordered combination of elements amounts to significantly more because it implements a specific parallel neural network architecture in which multiple RNNs process independent time series concurrently and merge learned outputs to generate a prediction, thereby improving computational efficiency and predictive performance relative to conventional serial architectures. This arrangement is not well-understood, routine, or conventional, as the specification describes parallel execution and output merging as a departure from prior serial RNN approaches that suffer from latency and inefficiency.
Examiner’s Response: Applicant's arguments have been fully considered but they are not persuasive. As explained in the Examiner’s response to the Applicant’s second arguments, the limitations of merging outputs and generating a prediction are abstract ideas, which alone cannot provide the technical improvement. In Step 2B, a parallel neural network architecture amounts to mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f).
Claim analysis under 101 as described in MPEP 2106.04 does not require the examiner to identify whether abstract ideas are well-understood, routine, and conventional. When an additional element has been identified as an insignificant extra-solution activity in Step 2A Prong 2, the analysis proceeds with evaluating whether the additional element is well-understood, routine, and conventional activity in Step 2B. The additional element “a parallel neural network architecture” was not evaluated as being well-understood, routine, and conventional activity because the Office Action identified it as mere instructions to apply the abstract ideas on a generic computer under MPEP 2106.05(f).
Claim 21 does not provide any details about how the parallel neural network architecture would improve computational efficiency and predictive performance relative to conventional serial architectures, or how the parallel execution would improve latency and inefficiency as argued in the remarks.
Applicant’s First Arguments Under 35 U.S.C. 103: On page 14, Applicant argues that the Examiner alleges that Abbaszadeh at paras. 96 and 97 teaches these features (claim 1, lines 11-15). However, this passage at best describes using GMM clustering to partition the operating space into regimes and support model construction and forecasting, not GMM clustering to determine a normalization value for the corresponding cluster normalization, let alone normalize the data value based on the normalization value as claimed.
Examiner’s Response: Applicant's arguments have been fully considered but they are not persuasive. Abbaszadeh was not relied upon for teaching the limitations in claim 1, lines 11-15 as argued in the remarks. Abbaszadeh was not relied upon for teaching determining a normalization value for the corresponding cluster or normalizing the data value based on the normalization value. Rather, Abbaszadeh’s paragraphs [0096] and [0097], lines 1-10 were relied upon for teaching the limitation “generate the mixture model, each cluster from among the plurality of clusters comprising a normal distribution”. Ouyang’s paragraphs [0080]-[0082] and reference claim 6, lines 1-7 were relied upon for teaching the limitation “identify a corresponding cluster, from among the plurality of clusters… , against which the data value is to be normalized, determine a normalization value for the corresponding cluster, and normalize the data value based on the normalization value”.
Applicant’s Second Arguments Under 35 U.S.C. 103: On page 14-15, Applicant argues that Ranjan’s time series data is a multivariate time-series data, in which multiple variables at each time step are processed together and are intentionally modeled as interdependent signals. The data values in Ranjan’s time series data do not vary independently, but rather are correlated and jointly modeled.
At para. 0047, Ranjan explains that the time-series data includes metrics such as CPU usage, memory usage, temperature, and disk usage, which “contribute[] to the performance of the server.” These variables are not independent; rather, they are physically and operationally correlated system metrics (e.g., CPU usage affects temperature and memory behavior), and are stored and analyzed together as a unified dataset.
Consistent with this, Ranjan’s LSTNet architecture is specifically designed to capture cross-variable and temporal dependencies, using convolutional layers to extract local dependencies across variables and recurrent and skip-recurrent layers to model long-term relationships across the combined signals. Because the model relies on learning relationships among variables within the same time series, the data values do not vary independently, but instead are jointly modeled and correlated.
In contrast, the claimed invention requires respective time series comprising sequential data values that vary independently, which are then processed separately by different RNNs operating in parallel. Ranjan does not disclose or suggest such independently varying sequences, nor does it teach separating independent time series across different neural networks. Instead, it teaches the opposite approach of aggregating interdependent variables into a single multivariate sequence and modeling their dependencies within a unified architecture.
Examiner’s Response: Applicant's arguments have been fully considered but they are not persuasive. Examiner treats the limitation “sequential data values that vary independently of one another over time” as data values that are different types of measurements. For example, a voltage would vary in units of volts, and a temperature would vary in units of degrees Fahrenheit. Since volts are different from degrees Fahrenheit, Ranjan’s data values of variables vary independently of one another over time.
Conclusion
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 Asher H. Jablon whose telephone number is (571)270-7648. The examiner can normally be reached Monday - Friday, 9:00 am - 6:00 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Abdullah Al Kawsar can be reached at (571)270-3169. 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.
/A.H.J./Examiner, Art Unit 2127
/ABDULLAH AL KAWSAR/Supervisory Patent Examiner, Art Unit 2127
Prosecution Insights