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 .
DETAILED ACTION
This action is in response to the application filed 07 November 2025. Claims 1, 10, 11, and 16 are amended. Claims 1-20 are pending and have been examined.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 07 November 2025 has been entered.
Response to Arguments
Applicant's arguments, see pages 10-12, filed 07 November 2025, with respect to the rejection of Claims 1-20 under 35 U.S.C. 101 have been fully considered but they are not persuasive.
APPLICANT'S ARGUMENT: Applicant argues (page 11, paragraph 5) that "The claims have been amended to emphasize the higher and lower dimensional data sets that enable the system to eliminate the need to convert the high dimension data of operation 410 above into the lower dimensioned data, which is an improvement to the efficiency of the computer system executing claimed methodology. Such an improvement is indicative that any alleged judicial exception has been integrated into a practical application."
EXAMINER'S RESPONSE: Examiner respectfully disagrees. In the rejection of amended Claim 1 under 35 U.S.C. 101 below, the claim is found to be directed to a mental process, and therefore ineligible. While the claim recites additional elements to the recited mental process steps, the claim does not appear to recite any additional elements integrating the claim into a practical application or providing significantly more. Examiner further notes that Applicant's argued improvement to computer efficiency does not appear to be recited or reflected in the rejected claims.
APPLICANT'S ARGUMENT: Applicant argues (page 12, paragraph 1) that "the claims now also recite 'receiving, via a computer network, operational parameter values from data generating devices connected to the computer network, and storing the operational parameter values.' As discussed during the interview, such an operation cannot be performed as a mental process."
EXAMINER'S RESPONSE: Examiner agrees that steps of receiving and storing operation parameter values are not accurately characterized as a mental process, but rather as additional elements of the claim. However, under Step 2A Prong 2 analysis of the Alice/Mayo framework, receiving and storing operational parameter values appear to amount to insignificant extra-solution activity. Under Step 2B analysis, receiving and storing operational parameter values appear to be well-understood, routine, conventional activities.
Applicant's arguments, see pages 12-13, filed 07 November 2025, with respect to the rejection of Claims 1-20 under 35 U.S.C. 103 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
APPLICANT'S ARGUMENT: Applicant argues (page 12, paragraph 5) that "none of the references alone or in combination teach, suggest, or disclose each claim limitation of the independent claims."
Applicant argues (page 13, paragraph 2) that "Guo does not disclose the use of a proxy model that labels data 'using a higher dimension set than the reduced dimension set' that was used to generate a first segmentation, as is recited in each of the independent claims."
EXAMINER'S RESPONSE: Applicant's arguments pertain in part to newly claimed matter. Further, Applicant's arguments are now moot. Claims 1-20 are rejected below under a new ground of rejection in light of Guo in view of Xie in view of Jain in view of Wenchel.
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 1-20 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1
Claim 1 is ineligible.
Step 1
Claim 1 recites a method, and thus the claimed process falls within a statutory category of invention.
Step 2A Prong 1
The claim recites generating, based on first segmentation parameters and a first set of data points from the operational parameter values, a first segmentation, wherein the first segmentation associates a segment with each data point in the first set of data points, and the first segmentation is based on a reduced dimension set of the first set of data points, which is a mental process. The claim recites labeling, based on the proxy model, a second set of data points from the operational parameter values to generate a labeled second set of data points, wherein the labeling is performed on the second set of data points from the operational parameter values using a higher dimension set than the reduced dimension set, which is a mental process. The claim recites generating, based on the first segmentation parameters, the first set of data points and the second set of data points, a second segmentation, wherein the second segmentation associates a segment with each data point in the first set of data points and the second set of data points, which is a mental process. The claim recites comparing the first segmentation and the second segmentation, which is a mental process. The claim recites triggering, based on the comparing, an adjustment of the first segmentation parameters to generate second segmentation parameters, which is a mental process. The claim recites generating, based on the adjustment, second segmentation parameters, which is a mental process. The claim recites generating, based on the second segmentation parameters, the first set of data points, the second set of data points, and a third set of data points from the operational parameter values, a third segmentation, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2
The additional element receiving, via a computer network, operational parameter values from data generating devices connected to the computer network amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element storing the operational parameter values amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering and outputting"). The additional element training a proxy model based on the first segmentation invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element outputting the labeled second set of data points amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element outputting a labeled fourth set of data points based on the third segmentation amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering").
Step 2B
The additional element receiving, via a computer network, operational parameter values from data generating devices connected to the computer network is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element storing the operational parameter values is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element training a proxy model based on the first segmentation invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element outputting the labeled second set of data points is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element outputting a labeled fourth set of data points based on the third segmentation is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 2
Claim 2 is ineligible.
Step 1
Regarding Claim 2, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites comparing the first segmentation and the second segmentation, which is a mental process. The claim recites determining whether each segment of the first segmentation has associated data points in the second segmentation, which is a mental process. The claim recites modifying the first segmentation parameters in response to at least one segment of the first segmentation lacking associated data points in the second segmentation, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 3
Claim 3 is ineligible.
Step 1
Regarding Claim 3, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites determining whether any segment of the second segmentation is associated with data points associated with two or more segments of the first segmentation, which is a mental process. The claim recites modifying the first segmentation parameters in response to a segment of the second segmentation being associated with data points associated with two or more segments of the first segmentation, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 4
Claim 4 is ineligible.
Step 1
Regarding Claim 4, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites comparing the labeled second set of data points to the second segmentation, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The additional element training the proxy model on the second segmentation in response to the comparing invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 5
Claim 5 is ineligible.
Step 1
Regarding Claim 5, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites determining a maximum distance between each data point of the first set of data points and the second set of data points, which is a mental process or a mathematical concept. The claim recites determining ... a centroid of a segment associated with a respective data point, which is a mental process or a mathematical concept. The claim recites modifying the first segmentation parameters based on the determining, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 6
Claim 6 is ineligible.
Step 1
Regarding Claim 6, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites modifying one or more of a distance function, density threshold, a number of expected clusters, a hyper-parameter, or a label definition of particular segments or clusters, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 7
Claim 7 is ineligible.
Step 1
Regarding Claim 7, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites reducing dimensions of the first set of data points to generate a reduced dimension set of data points, which is a mental process. The claim recites segmenting, based on the reduced dimension set of data points, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 8
Claim 8 is ineligible.
Step 1
Regarding Claim 8, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites generating, based on the second segmentation parameters, the first set of data points, a second set of data points, and a third set of data points, a third segmentation (as recited by Claim 1), wherein the third segmentation is based on at least one modified segmentation parameter of the first segmentation parameters, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 9
Claim 9 is ineligible.
Step 1
Regarding Claim 9, the rejection of Claim 8 is incorporated.
Step 2A Prong 1
The claim recites triggering, based on the comparing, an adjustment of the first segmentation parameters to generate second segmentation parameters, wherein the triggering is further based on receiving ... input to modify at least one of the first segmentation parameters (as recited by Claim 1), wherein modifying the first segmentation parameters includes modifying a normalization range for the first set of data points and the second set of data points, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 10
Claim 10 is ineligible.
Step 1
Regarding Claim 10, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites generating, based on first segmentation parameters and a first set of data points, a first segmentation (as recited in Claim 1), wherein the first set of data points represent electronic document data, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2
The additional element outputting labeled data to a network diagnostic application amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element outputting labeled data to a data filtering application amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering").
Step 2B
The additional element outputting labeled data to a network diagnostic application is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element outputting labeled data to a data filtering application is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 11
Claim 11 is ineligible.
Step 1
Claim 11 recites an apparatus comprising: a network interface configured to enable network communications; and one or more processors, and one or more memories storing instructions that when executed configure the one or more processors to perform operations, and thus the claimed apparatus falls within a statutory category of invention.
Step 2A Prong 1
The claim recites generating, based on first segmentation parameters and a first set of data points from the operational parameter values, a first segmentation, wherein the first segmentation associates a segment with each data point in the first set of data points, and the first segmentation is based on a reduced dimension set of the first set of data points, which is a mental process. The claim recites labeling, based on the proxy model, a second set of data points from the operational parameter values to generate a labeled second set of data points, wherein the labeling is performed on the second set of data points from the operational parameter values using a higher dimension set than the reduced dimension set, which is a mental process. The claim recites generating, based on the first segmentation parameters, the first set of data points and the second set of data points, a second segmentation, wherein the second segmentation associates a segment with each data point in the first set of data points and the second set of data points, which is a mental process. The claim recites comparing the first segmentation and the second segmentation, which is a mental process. The claim recites triggering, based on the comparing, an adjustment of the first segmentation parameters to generate second segmentation parameters, which is a mental process. The claim recites generating, based on the adjustment, second segmentation parameters, which is a mental process. The claim recites generating, based on the second segmentation parameters, the first set of data points, the second set of data points, and a third set of data points from the operational parameter values, a third segmentation, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2
The additional element receiving, via a computer network, operational parameter values from data generating devices connected to the computer network amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element storing the operational parameter values amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering and outputting"). The additional element training a proxy model based on the first segmentation invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element outputting the labeled second set of data points amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element outputting a labeled fourth set of data points based on the third segmentation amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering").
Step 2B
The additional element receiving, via a computer network, operational parameter values from data generating devices connected to the computer network is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element storing the operational parameter values is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element training a proxy model based on the first segmentation invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element outputting the labeled second set of data points is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element outputting a labeled fourth set of data points based on the third segmentation is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Claims 12-15, dependent on Claim 11, incorporate the rejection of Claim 11. Claims 12-15 incorporate substantively all the limitations of Claims 2-4 and 6, respectively, and are rejected under the same rationales.
Regarding Claim 16
Claim 16 is ineligible.
Step 1
Claim 16 recites a non-transitory computer readable storage medium comprising instructions that when executed configure one or more processors to perform operations, and thus the claimed manufacture falls within a statutory category of invention.
Step 2A Prong 1
The claim recites generating, based on first segmentation parameters and a first set of data points from the operational parameter values, a first segmentation, wherein the first segmentation associates a segment with each data point in the first set of data points, and the first segmentation is based on a reduced dimension set of the first set of data points, which is a mental process. The claim recites labeling, based on the proxy model, a second set of data points from the operational parameter values to generate a labeled second set of data points, wherein the labeling is performed on the second set of data points from the operational parameter values using a higher dimension set than the reduced dimension set, which is a mental process. The claim recites generating, based on the first segmentation parameters, the first set of data points and the second set of data points, a second segmentation, wherein the second segmentation associates a segment with each data point in the first set of data points and the second set of data points, which is a mental process. The claim recites comparing the first segmentation and the second segmentation, which is a mental process. The claim recites triggering, based on the comparing, an adjustment of the first segmentation parameters to generate second segmentation parameters, which is a mental process. The claim recites generating, based on the adjustment, second segmentation parameters, which is a mental process. The claim recites generating, based on the second segmentation parameters, the first set of data points, the second set of data points, and a third set of data points from the operational parameter values, a third segmentation, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2
The additional element receiving, via a computer network, operational parameter values from data generating devices connected to the computer network amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element storing the operational parameter values amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering and outputting"). The additional element training a proxy model based on the first segmentation invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element outputting the labeled second set of data points amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element outputting a labeled fourth set of data points based on the third segmentation amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering").
Step 2B
The additional element receiving, via a computer network, operational parameter values from data generating devices connected to the computer network is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element storing the operational parameter values is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element training a proxy model based on the first segmentation invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element outputting the labeled second set of data points is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network"). The additional element outputting a labeled fourth set of data points based on the third segmentation is well-understood, routine, conventional activity (see MPEP 2106.05(d), "receiving or transmitting data over a network").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Claims 17-20, dependent on Claim 16, incorporate the rejection of Claim 16. Claims 17-20 incorporate substantively all the limitations of Claims 2-5, respectively, and are rejected under the same rationales.
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.
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.
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.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Guo, et al., "Deep Clustering with Convolutional Autoencoders" (hereinafter "Guo") in view of Xie, et al., "Unsupervised Deep Embedding for Clustering Analysis" (hereinafter "Xie") in view of Jain, et al. (US 2021/0092036 A1, hereinafter "Jain") in view of Wenchel, et al. (U.S. 2021/0174258 A1, hereinafter "Wenchel").
Regarding Claim 1, Guo teaches:
a method (Guo, p. 378, 4.1 Data Sets: "The proposed DCEC [Deep Convolutional Embedded Clustering] method is evaluated on three image datasets") comprising:
receiving, via a computer network, ... values ... and storing the ... values (Guo, p. 378, 4.1 DataSets: "The proposed DCEC method is evaluated on three image datasets: MNISTfull ... MNIST-test ... USPS" and p. 379, 4.2 Experiment Setup: "the code is available at https://github.com/XifengGuo/DCEC," where source file README.md has command "bash ./download_usps.sh," where the download command receives and stores training and test data corresponding to the instant values);
generating ... a first segmentation (Guo, p. 378, 3.4 Optimization: "We first pretrain the parameters of CAE [Convolutional AutoEncoders] ... to get meaningful target distribution," where Guo's target distribution corresponds to the instant first segmentation), ... based on first segmentation parameters (Guo, p. 378, 3.4 Optimization: "We first pretrain the parameters of CAE ... to get meaningful target distribution," where Guo's pretrained parameters of the encoder correspond to the instant first segmentation parameters) and a first set of data points from the ... values (Guo, p. 379, 4.2 Experiment Setup: "Parameters setting. For SAE+k-means ..., the encoder network is set as a fully connected multilayer perceptron (MLP) ..., where d is the dimension of input data (features)," where Guo's input data corresponds to the instant first set), wherein the first segmentation associates a segment with each data point in the first set of data points (Guo, p. 377, 3.1 Structure of Deep Convolutional Embedded Clustering: "The clustering layer maps each embedded point
z
i
of input image
x
i
into a soft label," where Guo's soft label corresponds to the instant segment) ... ;
training a proxy model based on the first segmentation (Guo, p. 377, 3.1 Structure of Deep Convolutional Embedded Clustering: "The DCEC structure is composed of CAE (see Fig. 1) and a clustering layer which is connected to the embedded layer of CAE, as depicted in Fig. 2,"
where Guo's clustering layer corresponds to the instant proxy model);
labeling, based on the proxy model, a second set of data points from the ... values to generate a labeled second set of data points (Guo, p. 377, 3.2 Clustering Layer and Clustering Loss: "The clustering layer maintains cluster centers
{
μ
j
}
1
K
as trainable weights and maps each embedded point
z
i
into soft label
q
i
," where Guo's embedded point
z
i
corresponds to the instant second data point) ... ;
generating ... a second segmentation (Guo, p. 377, 3.2 Clustering Layer and Clustering Loss: "The clustering layer ... maps each embedded point
z
i
into soft label
q
i
," where Guo's clustering of embedded points correspond to the instant second segmentation), based on the first segmentation parameters, the first set of data points and the second set of data points (Guo, p. 378, 3.4 Optimization: "Update autoencoders' weights and cluster centers. As
∂
L
c
∂
z
t
and
∂
L
c
∂
μ
j
t
are easily derived according to [16], then the weights and centers can be updated by using backpropagation and mini-batch SGD straightforwardly," where Guo's weights and clusters correspond to the instant first segmentation parameters and second segmentation), wherein the second segmentation associates a segment with each data point in the first set of data points and the second set of data points (Guo, p. 377, 3.1 Structure of Deep Convolutional Embedded Clustering: "Then the clustering loss
L
c
is defined as Kullback-Leibler divergence (KL divergence) between the distribution of soft labels and the predefined target distribution," where Guo's clustering loss is calculated according to the target and soft distributions, which associate each data point into a cluster/segment);
comparing the first segmentation and the second segmentation (Guo, p. 377, 3.1 Structure of Deep Convolutional Embedded Clustering: "Then the clustering loss
L
c
is defined as Kullback-Leibler divergence (KL divergence) between the distribution of soft labels and the predefined target distribution," where Guo's target distribution and distribution of soft labels correspond to the instant first and second segmentation);
generating ... second segmentation parameters (Guo, p. 377, 3.2 Clustering Layer and Clustering Loss: "The clustering layer maintains cluster centers
{
μ
j
}
1
K
as trainable weights and maps each embedded point
z
i
into soft label
q
i
," where Guo's cluster centers correspond to the instant second segmentation parameters), based on the adjustment (Guo, p. 378, 3.4 Optimization: "Update autoencoders' weights and cluster centers. As
∂
L
c
∂
z
t
and
∂
L
c
∂
μ
j
t
are easily derived according to [16], then the weights and centers can be updated by using backpropagation and mini-batch SGD straightforwardly," where Guo's updated centers correspond to the instant adjusted second segmentation parameters);
generating, based on the second segmentation parameters, the first set of data points, the second set of data points, and a third set of data points from the ... values, a third segmentation (Guo, p. 378, 3.4 Optimization: "The training process terminates if the change of label assignments between two consecutive updates for target distribution is less than a threshold
δ
," where Guo's terminating label assignments corresponds to the instant third set of data points and Guo's terminating distribution of label assignments corresponds to the instant third segmentation).
Guo teaches generating, based on first segmentation parameters and a first set of data points from values, a first segmentation.
Guo does not explicitly teach the first segmentation is based on a reduced dimension set of the first set of data points and the labeling is performed ... using a higher dimension set than the reduced dimension set.
However, Xie teaches:
the first segmentation is based on a reduced dimension set of the first set of data points (Xie, p. 5, 4.3. Implementation: "During greedy layer-wise pretraining we initialize the weights to random numbers drawn from a zero-mean Gaussian distribution with a standard deviation of 0.01. Each layer is pretrained for 50000 iterations with a dropout rate of 20%. The entire deep autoencoder is further finetuned for 100000 iterations without dropout" where p. 4, 3.2. Parameter initialization: "A denoising autoencoder is a two layer neural network defined as [Eq. 6-9] where
D
r
o
p
o
u
t
⋅
... is a stochastic mapping that randomly sets a portion of its input dimensions to 0") and
the labeling is performed ... using a higher dimension set than the reduced dimension set (Xie, p. 5, 4.3. Implementation: "The entire deep autoencoder is further finetuned for 100000 iterations without dropout").
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Guo regarding generating, based on first segmentation parameters and a first set of data points from the operational parameter values, a first segmentation with those of Xie regarding the first segmentation is based on a reduced dimension set of the first set of data points and the labeling is performed using a higher dimension set than the reduced dimension set.
The motivation to do so would be to ensure that a trained model produces semantically meaningful and well-separated data representations (Xie, p. 3, 3.2. Parameter initialization: "We initialize DEC with a stacked autoencoder (SAE) because recent research has shown that they consistently produce semantically meaningful and well-separated representations on real-world datasets.... Thus the unsupervised representation learned by SAE naturally facilitates the learning of clustering representations with DEC. We initialize the SAE network layer by layer with each layer being a denoising autoencoder").
The Guo/Xie combination teaches receiving, via a computer network, values and storing the values.
The Guo/Xie combination does not explicitly teach operational parameter values from data generating devices connected to the computer network.
However, Jain teaches:
operational parameter values from data generating devices connected to the computer network (Jain, [0054]: "Based on the object-level estimates, the useful application-layer QoS metrics as described in Section 3 can be derived. ... FIG. 4 illustrates a functional block diagram of the network diagnostic performed by the controller 36 of the CPE 30 .... Specifically, FIG. 4 illustrates an object-level data obtaining function 36a, a network diagnostic function 36b, a traffic classification function 36c and a QoE estimation function 36d. In particular, the controller 36 performs the object-level data obtaining function").
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the Guo/Xie combination regarding receiving, via a computer network, values and storing the values with those of Jain regarding operational parameter values from data generating devices connected to the computer network.
The motivation to do so would be to facilitate performing network performance monitoring and root cause analysis (Jain, [0054]: "Based on the object-level estimates, the useful application-layer QoS metrics as described in Section 3 can be derived. The QoS metric estimates, in turn, can be used for performance monitoring, and network diagnostic to determine the root cause of degraded performance").
The Guo/Xie/Jain combination teaches generating first segmentation parameters, labeling a second set of data points, and generating a third set of data points from a third segmentation.
The Guo/Xie/Jain combination not explicitly teach outputting the labeled second set of data points, triggering ... an adjustment of the ... segmentation parameters ... based on the comparing ... to generate second segmentation parameters, and outputting a labeled ... set of data points based on the ... segmentation.
However, Wenchel teaches:
outputting the labeled second set of data points (Wenchel, Fig. 10, Step 1054: "Cause a machine learning (ML) model to generate intermediate output and model output based on the input data," and [0069]: "the system may cluster inputs or outputs based on computed feature importances, observe a goodness of fit metric for that clustering ( e.g., maximum radius or average distance), and/or make a determination to resample or take a ML model out of production," where Wenchel's clustered outputs correspond to the instant output labeled second set);
triggering ... an adjustment of the ... segmentation parameters (Wenchel, [0069]: "In some embodiments, actions can be triggered to improve model-level explanatory metrics. ... [I]f the entropy of explanatory weights of a large set of features is determined to be too high (e.g., above a predefined threshold) across a large set of inferences, it may be the case that the explainer is not working well and might benefit from additional data being incorporated into the base ML model," where Wenchel's incorporating additional data into the ML model corresponds to the instant adjusting segmentation parameters) ... based on the comparing ... to generate second segmentation parameters (Wenchel, [0069]: "Additional sampling may be triggered based on functions applied to the outputs of user-supplied explainers or system-supplied explainers. For example, in some embodiments, if the entropy of explanatory weights of a large set of features is determined to be too high (e.g., above a predefined threshold) across a large set of inferences," where Wenchel's explanatory weights correspond to the instant second segmentation parameters, and determined too high corresponds to based on the comparing);
outputting a labeled ... set of data points based on the ... segmentation (Wenchel, [0064]: "The second service includes providing explanatory metrics to the system. ... These explanatory metrics may be real-valued weights associated with: individual features, sets of features, the models themselves (optionally with timestamps), and/or associated inference input/output pairs from the user's ML model(s)," where Wenchel's inference output corresponds to the instant labeled data points).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the Guo/Xie/Jain combination regarding generating first segmentation parameters, labeling a second set of data points, and generating a third set of data points from a third segmentation with those of Wenchel regarding outputting the labeled second set of data points, triggering an adjustment of the segmentation parameters based on the comparing to generate second segmentation parameters, and outputting a labeled set of data points based on the segmentation.
The motivation to do so would be to facilitate model explainability for humans in light of changes in data processed by ML systems (Wenchel, [0016]: "Tradeoffs exist between the complexity of known ML model ..., the human interpretability of the models, the amount of data used to initialize or 'train' the ML models.... Making matters more complex, the sample data used to train a ML model may change over time" and [0017]: "Some embodiments described herein provide a holistic method for proactively monitoring and measuring the overall health of a ML-based computer system.... Such methods may include measurement and human-interpretable display of the health and performance of the ML-based computer systems").
Regarding Claim 11, the Guo teaches:
an apparatus comprising ... one or more processors, and one or more memories storing instructions that when executed configure the one or more processors (Guo, p. 379, 4.2 Experiment Setup: "Our implementation is based on Python and Keras [2] and the code is available at https://github.com/XifengGuo/DCEC," where a computing apparatus comprising a processor and memory storing instructions is inherent in Guo's implementation) a network interface configured to enable network communications (Guo, https://github.com/XifengGuo/DCEC, source file README.md, command "sudo pip install keras scikit-learn," where the pip tool is understood under BRI to perform requests over a communications network) to perform operations comprising precisely those steps recited by the method of Claim 1.
Regarding Claim 16, Guo teaches:
a non-transitory computer readable storage medium (Guo, https://github.com/XifengGuo/DCEC, source file README.md, command "sudo pip install keras scikit-learn," where the pip tool is understood under BRI to store executable source code to a non-transitory computer-readable store medium) comprising instructions that when executed configure one or more processors (Guo, p. 379, 4.2 Experiment Setup: "Our implementation is based on Python and Keras [2] and the code is available at https://github.com/XifengGuo/DCEC," where instructions executable to configure a processor are inherent in the source code listing) to perform operations comprising precisely those steps recited by the method of Claim 1.
Regarding Claim 2, the rejection of Claim 1 is incorporated. The Guo/Xie/Jain/Wenchel combination teaches:
comparing the first segmentation and the second segmentation comprises determining whether each segment of the first segmentation has associated data points in the second segmentation (Guo, p. 375, 2 Convolutional AutoEncoders: "The key factor of the proposed CAE is the aggressive constraint on the dimension of embedded layer. ... Learning such under-complete representations forces the autoencoder to capture the most salient features of the data. Thus we force the dimension of embedded space to equal to the number of clusters of dataset")
(Guo, p. 377, 3.1 Structure of Deep Convolutional Embedded Clustering: "Then the clustering loss Lc is defined as Kullback-Leibler divergence (KL divergence) between the distribution of soft labels and the predefined target distribution"); and
modifying the first segmentation parameters in response to at least one segment of the first segmentation lacking associated data points in the second segmentation (Guo, p. 378, 3.4 Optimization: "Update autoencoders' weights and cluster centers. As
∂
L
c
∂
z
t
and
∂
L
c
∂
μ
j
t
are easily derived according to [16], then the weights and centers can be updated by using backpropagation and mini-batch SGD straightforwardly," where Guo's updated weights correspond to the instant modified first segmentation parameters, and where Guo's clustering loss (
L
c
) corresponds to lacking association between segmentations).
Claims 12 and 17 incorporate substantively all the limitations of claim 2 in device and product forms, respectively, and are rejected under the same rationale.
Regarding Claim 3, the rejection of Claim 1 is incorporated. The Guo/Xie/Jain/Wenchel combination teaches:
wherein comparing the first segmentation and the second segmentation comprises determining whether any segment of the second segmentation is associated with data points associated with two or more segments of the first segmentation (Guo, 3.1 Structure of Deep Convolutional Embedded Clustering: "Then the clustering loss
L
c
is defined as Kullback-Leibler divergence (KL divergence) between the distribution of soft labels and the predefined target distribution," where Guo's target distribution and distribution of soft labels correspond to the instant first and second segmentation, and where all data points of the second segmentation are associated with all segments of the first segmentation due to Guo's probability distribution of label assignment, rather than a one-to-one assignment) and modifying the first segmentation parameters in response to a segment of the second segmentation being associated with data points associated with two or more segments of the first segmentation (Guo, p. 378, 3.4 Optimization: "We first pretrain the parameters of CAE.... After pretraining, the cluster centers are initialized .... Then set
γ
=
0.1
and update CAE’s weights, cluster centers and target distribution
P
as follows," where the hyperparameter
γ
causes the clustering loss of cluster assignments to be accounted for, and where Guo's pretrained parameters of the encoder correspond to the instant first segmentation parameters).
Claims 13 and 18 incorporate substantively all the limitations of claim 3 in device and product forms, respectively, and are rejected under the same rationale.
Regarding Claim 4, the rejection of Claim 1 is incorporated. The Guo/Xie/Jain/Wenchel combination teaches:
comparing the labeled second set of data points to the second segmentation (Guo, p. 378, 3.4 Optimization: "Update autoencoders' weights and cluster centers. As
∂
L
c
∂
z
t
and
∂
L
c
∂
μ
j
t
are easily derived according to [16], then the weights and centers can be updated by using backpropagation and mini-batch SGD straightforwardly," where Guo's updating centers according to clustering loss comprises the instant comparing), and training the proxy model on the second segmentation in response to the comparing (Guo, ibid., with "The training process terminates if the change of label assignments between two consecutive updates for target distribution is less than a threshold
δ
," where Guo's training of the clustering layer continues based in part on the clustering loss).
Claims 14 and 19 incorporate substantively all the limitations of claim 4 in device and product forms, respectively, and are rejected under the same rationale.
Regarding Claim 5, the rejection of Claim 1 is incorporated. The Guo/Xie/Jain/Wenchel combination teaches:
determining a ... distance between each data point of the first set of data points ... and a centroid of a segment associated with a respective data point (Guo, p. 378, 3.4 Optimization: "We first pretrain the parameters of CAE.... After pretraining, the cluster centers are initialized by performing
k
-means on embedded features of all images," where Guo's performing k-means corresponds to the instant determining a distance between each point of the first set) and determining a ... distance between each data point of... the second set of data points ... and a centroid of a segment associated with a respective data point (Guo, p. 377, 3.2 Clustering Layer and Clustering Loss: "The clustering layer maintains cluster centers
{
μ
j
}
1
K
as trainable weights and maps each embedded point
z
i
into soft label
q
i
," where Guo's embedded point
z
i
is associated a cluster center according to the probability of
z
i
belonging to the cluster, according to Eq. 6), and modifying the first segmentation parameters based on the determining (Guo, p. 378, 3.4 Optimization: "After pretraining, the cluster centers are initialized by performing
k
-means on embedded features of all images. Then set
γ
=
0.1
and update CAE’s weights, cluster centers and target distribution
P
as follows," where Guo's pretrained parameters of the encoder correspond to the instant first segmentation parameters, which are updated according to the determined centers after pre-training).
Wenchel further teaches:
determining a maximum distance between each data point of ... data points and a centroid of a segment associated with a respective data point (Wenchel, [0069]: In other embodiments, the system may cluster inputs or outputs based on computed feature importances, observe a goodness of fit metric for that clustering ( e.g., maximum radius or average distance), and/or make a determination to resample or take a ML model out of production," where Wenchel's maximum cluster radius corresponds to the instant maximum distance).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the Guo/Xie/Jain/Wenchel combination regarding determining a centroid of a segment associated with a respective data point and modifying the first segmentation parameters based on the determining with the further teachings of Wenchel regarding determining a maximum distance between each data point of data points and a centroid of a segment associated with a respective data point.
The motivation to do so would be to provide a potentially actionable metric for monitoring or improving ML model performance (Wenchel, [0069]: "In some embodiments, actions can be triggered to improve model-level explanatory metrics. ... In other embodiments, the system may cluster inputs or outputs based on computed feature importances, observe a goodness of fit metric for that clustering ( e.g., maximum radius or average distance), and/or make a determination to resample or take a ML model out of production").
Claim 20 incorporates substantively all the limitations of claim 5 in product form and is rejected under the same rationale.
Regarding Claim 6, the rejection of Claim 1 is incorporated. The Guo/Xie/Jain/Wenchel combination teaches:
wherein modifying the first segmentation parameters includes modifying one or more of ... a hyper-parameter ... (Gau, p. 378, 3.4 Optimization: "We first pretrain the parameters of CAE by setting
γ
=
0
to get meaningful target distribution. After pretraining, the cluster centers are initialized by performing
k
-means on embedded features of all images. Then set
γ
=
0.1
and update CAE’s weights, cluster centers and target distribution P as follows," where Gau's updating CAE weights (i.e., first segmentation parameters) includes modifying the hyperparameter
γ
).
Claim 15 incorporates substantively all the limitations of claim 6 in device form and is rejected under the same rationale.
Regarding Claim 7, the rejection of Claim 1 is incorporated. The Guo/Xie/Jain/Wenchel combination teaches:
wherein the first segmentation comprises reducing dimensions of the first set of data points to generate a reduced dimension set of data points (Guo, p. 375, 2 Convolutional AutoEncoders: "The intuitive way of avoiding identity mapping is to control the dimension of latent code h lower than input data x"), and segmenting, based on the reduced dimension set of data points (Guo, p. 377, 3.1 Structure of Deep Convolutional Embedded Clustering: "The clustering layer maps each embedded point zi of input image xi into a soft label").
Regarding Claim 8, the rejection of Claim 1 is incorporated. The Guo/Xie/Jain/Wenchel combination teaches:
wherein the third segmentation is based on at least one modified segmentation parameter of the first segmentation parameters (Guo, p. 378, 3.4 Optimization: "We first pretrain the parameters of CAE ... to get meaningful target distribution. ... [T]he weights and centers can be updated by using backpropagation and mini-batch SGD .... The training process terminates if the change of label assignments between two consecutive updates for target distribution is less than a threshold
δ
," where Guo's terminating label assignments corresponds to the instant third segmentation and Guo's parameters of CAE updated using backpropagation corresponds to the instant modified segmentation parameter).
Regarding Claim 9, the rejection of Claim 1 is incorporated. The Guo/Xie/Jain/Wenchel combination teaches:
modifying the first segmentation parameters includes modifying a normalization range for the first set of data points and the second set of data points (Guo, p. 377, 3.2 Clustering Layer and Clustering Loss: "The clustering layer maintains cluster centers
{
μ
j
}
1
K
as trainable weights and maps each embedded point
z
i
into soft label
q
i
by Student's t-distribution [9]:
q
i
j
=
1
+
z
i
-
μ
i
2
-
1
∑
j
1
+
z
i
-
μ
2
-
1
(6)
where
q
i
j
is the jth entry of
q
i
, representing the probability of
z
i
belonging to cluster
j
. ¶ The clustering loss is defined as:
L
c
=
K
L
(
P
|
Q
=
∑
i
∑
j
p
i
j
log
q
i
j
p
i
j
(7)
where
P
is the target distribution" and p. 6, 3.4 Optimization, Update autoencoders' weights and cluster centers: "As
∂
L
c
∂
z
i
and
∂
L
c
∂
μ
j
are easily derived according to [16], then the weights and centers can be updated by using backpropagation and mini-batch SGD straightforwardly," where Guo's normalized Euclidean distance from embedded points
z
i
to cluster centers
μ
i
, calculated for backpropagation, corresponds to the instant normalization range).
Regarding Claim 10, the rejection of Claim 1 is incorporated. Jain further teaches:
wherein the outputting comprises outputting labeled data (Jain, [0123]: "QoS metrics and network stats can be utilized for designing automated ... network diagnostics algorithms. ... This can be framed as a machine-learning classification (supervised learning) problem, where a machine-learning model is trained using historical notes from network operators (ground-truth labels) .... [T]he controller 36 (or the controller 24) can perform the network diagnostic of the network condition of the TCP connection by determining a root cause of the network condition of the network connection" where root causes are labeled, as in [0124]: "accurate labels for the root causes") to a network diagnostic application (Jain, Fig. 1, and [0039]: "with the network diagnostic algorithm of the present disclosure, such application-layer QoS metrics are estimated from within the ISP's network, without access to the client device. FIG. 1 below shows one particular application in a satellite network setting") or the first set of data points represent electronic document data, and the outputting comprises outputting labeled data to a data filtering application (Jain teaches the first alternative).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the Guo/Xie/Jain/Wenchel combination regarding outputting the labeled second set of data points and outputting a labeled fourth set of data points based on the third segmentation with the further teachings of Jain regarding the outputting comprises outputting labeled data to a network diagnostic application.
The motivation to do so would be to facilitate providing proactive root-cause diagnostics and resolution for scenarios involving network performance degradation (Jain, [0047]: "FIG. 2 shows the main use case/motivation for the network diagnostic algorithm and the network diagnostic apparatus of the present disclosure. In particular, the passive monitoring of the application-layer QoS metrics (and QoE metrics) enable proactively detecting any performance degradation, and the automated network diagnostics algorithm can give the root causes, which can be used to troubleshoot and fix the identified issues").
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ROBERT N DAY whose telephone number is (703)756-1519. The examiner can normally be reached M-F 9-5.
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, Kakali Chaki can be reached at (571) 272-3719. 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.
/R.N.D./Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122