DETAILED ACTION
This Office Action is sent in response to the Applicant’s Communication received on 04/13/2026 for application number 18/209,024. The Office hereby acknowledges receipt of the following and placed of record in file: Specification, Drawings, Abstract, Oath/Declaration, IDS, and Claims.
Claims 1, 5, 11, and 15 are amended.
Claim 21 is new.
Claims 1-21 are pending.
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 .
Response to Arguments
35 USC 103
On page 8 of the remarks section, Applicant argues that Yin is mischaracterized as Yin teaches "count the number of times each feature value" generates one unidimensional histogram per feature. Histogram(s) are not the claimed "multidimensional point".
The Examiner respectfully disagrees. The alleged “histograms” were not cited as teaching the claimed “multidimensional point”. Rather, Yin’s deduplicated dataset was cited as teaching the claimed “multidimensional point”.
On page 9 of the remarks section, Applicant argues that Lin is mischaracterized for citing “weight” as teaching the claimed “error”.
The Examiner respectfully disagrees. The Office Action did cite “weights” alone as teaching the claimed “scaled error”. To clarify, the Office Action cites “the loss minimization engine 312 can minimize a loss function (e.g., including an error function, such as mean squared error, and the penalty applied by the penalty engine 314) to maintain the outputs of the original set of filters as weights, scaling factors, and/or other parameters of the duplicate filters are updated and as the candidate filters or parameters (e.g., weights) of the candidate filters are removed from the duplicate set of filters” (Yin, para 0060) as teaching the claimed “scaled error”.
On page 9 of the remarks section, the Applicant argues that Lin is mischaracterized for citing "minimization ... minimize" as teaching the claimed "increasing" as minimizing is the opposite of increasing.
The Examiner finds the Applicant’s argument persuasive. However, after further review and search, a new ground of rejection is provided.
On pages 9 and 10 of the remarks section, Applicant argues that Lin teaches a neural network while Yin lacks machine learning, and the Office action says "Yin does not teach ... increasing ... accuracy of the reconstructive model." The Office action says "Yin ... by a ... model. .. Para 0032, The deduplication module"; however that "module" is not a machine learning model. The Office action says "modified Yin' s teachings to incorporate the teachings of Lin and provide a scaled error in order to tune model weights for improved accuracy." Here, "tune model weights" is Lin's neural weights because Yin lacks weights. Here, "provide a scaled error" is not taught by Yin, and the Office action says "Yin does not teach ... a scaled error". Thus, "in order to tune model weights for improved accuracy" means improving Lin's neural accuracy by Lin's tuning and by Lin's supposed error scaling. In other words, "in order to tune model weights for improved accuracy" means Lin improving Lin itself, which is not a valid combination rationale. Under KSR: there is no motivation to combine Lin; combining Lin is illegal; and prima facie obviousness is unestablished.
The Examiner respectfully disagrees. The Examiner is not suggesting a bodily incorporation of Lin to Yin, but rather that the methodology of Lin’s increasing, based on increasing frequency of distinct point, a scaled error of reconstruction of the distinct point to the teachings of Yin to tune model weights for improved accuracy would have been an obvious motivation to combine in order to arrive at the claimed limitation. In response to applicant's argument, the test for obviousness is not whether the features of a secondary reference may be bodily incorporated into the structure of the primary reference; nor is it that the claimed invention must be expressly suggested in any one or all of the references. Rather, the test is what the combined teachings of the references would have suggested to those of ordinary skill in the art. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981). However, upon further review and search, a new ground of rejection has been proposed in this Office Action, necessitated by amendment.
On page 10 of the remarks section, Applicant argues that amended claim 5 has additional distinctiveness over reference Del Monte.
While the Applicant’s arguments have been considered, the argument is moot since the new rejection relies on a different set of references for teaching the newly amended claim.
On page 10 of the remarks section, Applicant argues that although the Office action says "Armangau ... relate to textual data manipulation", mischaracterized Armangau lacks text.
The Examiner finds the Applicant’s argument persuasive. However, after further review and search, a new ground of rejection is provided.
On pages 10-11 of the remarks section, the Applicant argues that Claim 7 depends from Claim 4 for which the Office action says "Zhang ... Para 0022, the query text in the initial training samples of the preset text matching model, i.e. the keywords input into the preset text matching model, is clustered. Then, the clustered query text is deduplicated and corrected according to". Claim 7 recites "accuracy of the reconstructive model". For Claim 7, the Office action quotes "Kascenas ... At test time" and "Figure 1 ... noise is added to the foreground of the healthy image, and the network is trained to reconstruct the original image. At test time (bottom), the pixelwise post processed reconstruction error is used as". It is unclear whether the claimed model processes the cited "image ... pixel wise" or instead processes cited "i.e. the keywords input into the preset text matching model". Due to ambiguity under KSR: there is no motivation to combine Kascenas; combining Kascenas is illegal; and prima facie obviousness is unestablished.
The Examiner respectfully disagrees. The Examiner is not suggesting a bodily incorporation of Kascenas to Zhang, but rather that the methodology of Kascenas’s anomaly detection accuracy to the teachings of Zhang to improve threshold for machine learning systems would have been an obvious motivation to combine in order to arrive at the claimed limitation. The test for obviousness is not whether the features of a secondary reference may be bodily incorporated into the structure of the primary reference; nor is it that the claimed invention must be expressly suggested in any one or all of the references. Rather, the test is what the combined teachings of the references would have suggested to those of ordinary skill in the art. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981).
On page 11 of the remarks section, Applicant argues that Claim 7 directly depends from Claim 4 for which the Office action says "Zhang ... Para 0022 ... the labels corresponding to the samples". Here, "the labels" is supervision. Likewise, Claim 7 indirectly depends from Claim 1 for which the Office action says "Lin ... Para 0060 ... as supervision". For Claim 7, the Office action instead says "Kascenas et al. (Denoising
Autoencoders for Unsupervised". It is unclear whether the cited machine learning is
"Unsupervised" or instead "the labels" are "as supervision". Due to ambiguity under KSR: there is no motivation to combine Kascenas; combining Kascenas is illegal; and prima facie obviousness is unestablished.
The Examiner respectfully disagrees. The Examiner is not suggesting a bodily incorporation of Kascenas to Lin, but rather that the methodology of Kascenas’s anomaly detection accuracy to the teachings of Lin to improve threshold for machine learning systems would have been an obvious motivation to combine in order to arrive at the claimed limitation. The test for obviousness is not whether the features of a secondary reference may be bodily incorporated into the structure of the primary reference; nor is it that the claimed invention must be expressly suggested in any one or all of the references. Rather, the test is what the combined teachings of the references would have suggested to those of ordinary skill in the art. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981).
New Claim 21 is not neural per the specification (0041). As discussed above for Claim 1, Lin instead is neural and, thus, incompatible.
The Examiner respectfully disagrees. The teachings newly introduced reference Consoli are combined with the teachings of primary reference Yin and not secondary reference Lin.
Therefore, the 35 USC 103 rejection is maintained.
35 USC 101
On page 12 of the remarks section, the Applicant argues that Claim 1 in its present form has unconventional step "increasing, based on said increasing said observed frequency of the distinct multidimensional point, a scaled error of the reconstruction of the distinct multidimensional point". Here, "increasing, based on... , a scaled error of the reconstruction" is an unconventional increase of error. In that way, the claimed "scaled error" is unconventionally generated. That is, the claimed "increasing ... scaled error" is an unconventional number. The Office action does not mention conventionality.
The Examiner respectfully disagrees. The claimed limitation "increasing, based on said increasing said observed frequency of the distinct multidimensional point, a scaled error of the reconstruction of the distinct multidimensional point" was analyzed under Step 2A Prong One as being an abstract idea. The action of increasing a scaled error is a limitation recited at a high-level generality and, under the Broadest Reasonable Interpretation (BRI), can be interpreted as a procedure of human observation, evaluation, judgement, or opinion. There are no explicit steps or details recited that meaningfully limit the claimed limitation to falling outside the grouping of abstract ideas. The claimed limitation can be performed mentally with the aid of pen and paper, and is therefore a mental process.
On pages 12 and 13 of the remarks section, Applicant argues that new step "training ... including based on the scaled error of the reconstruction" is training based on unconventional reconstruction loss. In that way, Claim 1 in its present form has unconventional training by unconventional loss. Unconventional loss is not a training input from a training corpus. Unconventional loss innovates the training step itself. New claim step "training" is nonabstract.
The Examiner respectfully disagrees. The claimed limitation “training the reconstructive model, including based on the scaled error of the reconstruction of the distinct multidimensional point, increasing accuracy of the reconstructive model” was analyzed under Step 2A Prong Two as an additional element limiting the judicial exception of increasing a scaled error to the particular field of use or technological environment of reconstructive modeling. Therefore, the cited additional element does not integrate the judicial exception into a practical application and cannot provide for an inventive concept.
On page 13 of the remarks section, the Applicant argues that the claimed "scaled error" is an unconventionally generated number, and the specification (Overview) teaches "By a novel objective function that uses a count of duplicates as a scaling factor, this approach has at least the following advantages." Elsewhere the Office action says "provide a scaled error in order to tune model weights for improved accuracy" (for Claims 1-2), "to give improved reference data" (for Claim 3), "to obtain higher quality sample data" (for Claims 4 and 7), "to standardize data and enhance data efficiency" (for Claim 5), "to support more efficient computations" (for Claim 6), "provide anomaly detection accuracy ... in order to improve intensity threshold" (for Claim 7), "to provide a comprehensive training set to improve machine learning models" (for Claim 8), "to enhance machine learning systems by implementing a processing methodology that improves efficiency and scalability" (for Claim 9), and "to improve model robustness" (for Claim 10). Claim 1 in its present form improves technology performance characteristics (i.e. accuracy and, as discussed below, time and space). The section 101 rejection does not consider any improvement.
The Examiner respectfully disagrees. First, the cited usage of the “scaled error” is not being used in a specific manner to be considered as unconventional. Additionally, the conventional usage of the “scaled error” is established in the 35 USC 103 rejection. Second, only claims 1 and 2 are directly related to the scaled error and, as mentioned above, the claimed limitation uses the “scaled error” in a generic manner. Claims 3-10 are not directly related to the scaled error and therefore provide motivations for differing limitations of the claimed invention. Third, claim 1 in its present form does not improve the technology because the limitations are either directed to judicial exceptions or additional elements that do not integrate the exception into a practical application.
On pages 13 and 14 of the remarks section, Applicant argues that, unlike cited "iteratively ... various iterations", the specification (Overview) instead teaches "no additional iteration" as "faster model training time. Training acceleration is due to the fact that no additional iteration needs to be performed for duplicates". Likewise, the specification teaches "Efficient" (Title), "accelerated training" (Field), "frequency-based error scaling herein accelerates training" (1.4 Point Frequency As Novel Scaling Factor For Error), and "By a novel objective function that uses a count of duplicates as a scaling factor, this approach has at least the following advantages .... Two orders of magnitude speedup of all ML pipeline stages (e.g. corpus preparation, corpus information theoretics, or model training) that are downstream of deduplication. No feature extraction from duplicates for acceleration" (Overview). Here, "novel. .. speedup ... training" characterizes Claim 1 in its present form. Likewise, the specification (0062) teaches "Training speedup and feature-vector generation speedup are each l00x (i.e. 10,000 percent) ... for the deduplicated training corpus."
The Examiner respectfully disagrees. The Applicant not shown any nexus between the argument and any particular limitation or groups of limitations in the claimed invention. The argument merely makes conclusory statements regarding improvements without particularly linking them to any particular claim limitation.
On page 14 of the remarks section, Applicant argues that the claimed limitation "encoding ... from the plurality of distinct multidimensional points" is encoding less than a whole training corpus such as the claimed "original plurality" and that Claim 1 in its present form trains with unconventionally less than a whole training corpus, which is unconventionally less training, which is unconventionally accelerated training.
The Examiner respectfully disagrees. Although “encoding less than a whole training corpus” appears to be disclosed invention, the claimed limitation does not explicitly recite this language. If the specification sets forth an improvement in technology, the claim must be evaluated to ensure that the claim itself reflects the disclosed improvement. That is, the claim includes the components or steps of the invention that provide the improvement described in the specification. See MPEP 2106.04(d)(1).
On page 15 of the remarks section, the Applicant argues that the specification (Overview) teaches "novel... space savings" and "By a novel objective function that uses a count of duplicates as a scaling factor, this approach has at least the following advantages. Order(s) of magnitude space savings by corpus deduplication." Likewise,
the specification (1.2 Deduplicated Training Corpus) teaches "the deduplicated training corpus is much smaller than original training corpus 120, which conserves space in a training computer that uses the deduplicated training corpus without accessing original training corpus 120." Claim 1 in its present form has new step "encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points". Here, "encoding ... from the plurality of distinct multidimensional points" is encoding less than a whole training corpus such as the claimed "original plurality". That is, Claim 1 in its present form trains with unconventionally less than a whole training corpus, which is unconventionally less data, which is unconventional conservation of space (i.e. memory). Likewise, the specification teaches "The approach herein has a smaller memory footprint" (Overview) and "The deduplicated training corpus is one percent of the size (i.e. count of multidimensional points) of original training corpus 120 ... Training ... and feature-vector generation ... require 99 percent less space for the deduplicated training corpus" (0062). Claim 1 in its present form needs unconventionally less memory in training.
The Examiner respectfully disagrees. First, in the case of claim 1 in its present form, the claimed limitations are written with a high level of generality lacking the specific steps to be considered unconventional. Second although “encoding less than a whole training corpus”, “The deduplicated training corpus is one percent of the size (i.e. count of multidimensional points) of original training corpus”, and “Training ... and feature-vector generation ... require 99 percent less space for the deduplicated training corpus" appear to be disclosed invention, the claimed limitation does not explicitly recite this language. If the specification sets forth an improvement in technology, the claim must be evaluated to ensure that the claim itself reflects the disclosed improvement. That is, the claim includes the components or steps of the invention that provide the improvement described in the specification. See MPEP 2106.04(d)(1).
On pages 16-17 of the remarks section, Applicant argues that the state of the art has minority oversampling that is inversely proportional to minority frequency. Thus the rarer is a minority, the more important is the minority and the more biased is the sampling toward the minority. In other words, rebalancing works against a natural training bias towards learning the most common (i.e. majority) data. Decreased accuracy from rebalancing is worst for unsupervised training" (Background). The specification teaches unconventional "maintaining accuracy" (Field), "Based on the scaled error of the reconstruction of the particular distinct multidimensional point, accuracy of the reconstructive ML model is increased" (Overview), and "bias is avoided by the frequency-based error scaling as discussed earlier herein, which may increase the accuracy of reconstructive model 110 beyond the state of the art for a given training budget (i.e. time, iterations, multidimensional points, or batches)" (0033).
The Examiner respectfully disagrees. The Applicant not shown any nexus between the argument and any particular limitation or groups of limitations in the claimed invention. The argument merely makes conclusory statements regarding improvements without particularly linking them to any particular claim limitation.
On page 17 of the remarks section, Applicant argues that Claim 1 in its present form has performance enhancing step "based on the scaled error of the reconstruction ... increasing accuracy of the reconstructive model". Likewise elsewhere for Claims 1-2 the Office action says "provide a scaled error in order to tune model weights for improved accuracy". Claim 1 in its present form unconventionally increases accuracy of
internal operation of the claimed training computer.
The Examiner respectfully disagrees. As mentioned above, claim 1 in its current recites limitations with a high level of generality that are either directed to judicial exceptions or additional elements that do not integrate the judicial exceptions into a practical application. Moreover, the cited “increasing accuracy of the reconstructive model” is a bare assertion of an improvement that lacks the necessary detail to be considered unconventional.
On page 18 of the remarks section, Applicant argues that those two nonabstract steps "increasing accuracy" and "training" cooperating with that unconventional step "increasing ... a scaled error of the reconstruction" are three meaningfully ordered steps that unconventionally improve performance of internal operation of the claimed training computer, including unconventional acceleration of training, unconventional conservation of space (i.e. memory), and unconventionally increased accuracy. In those several ways, Claim 1 in its present form has unconventionally improved training that is significantly more than an abstraction.
The Examiner respectfully disagrees. The aforementioned three steps are all written at a high level of generality that all lack necessary detail to be considered unconventional. Both of the “increasing” steps were analyzed under Step 2A Prong One as being considered mental processes. There is insufficient steps and details that meaningfully limit these actions to falling outside the boundaries of mental processes. The training step was analyzed under Step 2A Prong Two as an additional element that merely limits the judicial exceptions to the field or technological environment of reconstructive modeling in machine learning, which cannot provide for an inventive concept.
On pages 18 and 19 of the remarks section, the Applicant argues that Unconventional step "increasing ... a scaled error of the reconstruction" improves the
performance of the "training" step. Unconventional step "increasing ... a scaled error of the reconstruction" and claimed "by a reconstructive model" in Claim 1 in its present form would not be characterized as "merely indicating a field of use or technological environment". Claim 1 in its present form would not be characterized as "mentally with the aid of pen and paper".
The Examiner respectfully disagrees. As mentioned above, the claimed “increasing ... a scaled error of the reconstruction" was determined to be a mental process for reciting steps with a level of generality. Additionally, merely mentioning that the judicial exception is performed "by a reconstructive model" does not limit the judicial exception to falling outside the bounds of a mental process, nor does it provide for practical application.
Therefore, the 35 USC 101 rejection is maintained.
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 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claims 1-10 are directed to a method. Claims 11-20 are directed to a non-transitory computer-readable media. Therefore, all claims are directed to one of the four categories of patent eligible subject matter.
Claim 1
Step 2A Prong 1:
Claim 1 recites:
“A method comprising: detecting a plurality of distinct multidimensional points in an original plurality of multidimensional points that contains duplicates;” Detecting a plurality of distinct multidimensional points is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
“increasing, based on duplicates in the original plurality of multidimensional points, a respective observed frequency of each distinct multidimensional point in the plurality of distinct multidimensional points;” Increasing a respective observed frequency is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
“generating, from the feature vector, a reconstruction of a distinct multidimensional point [by a reconstructive model];” Generating a reconstruction of the distinct multidimensional point is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
“increasing, based on said increasing said observed frequency of the distinct multidimensional point, a scaled error of the reconstruction of the distinct multidimensional point;” Increasing a scaled error of the reconstruction is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
“wherein the method is performed by one or more computers;” Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)).
“by a reconstructive model;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
“training the reconstructive model, including based on the scaled error of the reconstruction of the distinct multidimensional point, increasing accuracy of the reconstructive model;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
“wherein the method is performed by one or more computers;” Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)) which cannot provide an inventive concept.
“by a reconstructive model;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
“training the reconstructive model, including based on the scaled error of the reconstruction of the distinct multidimensional point, increasing accuracy of the reconstructive model;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 2
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“training the reconstructive model with a training corpus that consists of the plurality of distinct multidimensional points;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“training the reconstructive model with a training corpus that consists of the plurality of distinct multidimensional points;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 3
Step 2A Prong 1:
Claim 3 recites:
“generating a batch that represents more multidimensional points than the batch contains;” Generating a batch that represents more multidimensional points than the batch contains is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
Step 2A Prong Two and Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B. Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2. The claim is ineligible.
Claim 4
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“wherein each multidimensional point in the original plurality of multidimensional points represents a respective textual command;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“wherein each multidimensional point in the original plurality of multidimensional points represents a respective textual command;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 5
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“wherein said detecting the plurality of distinct multidimensional points comprises normalization of whitespace;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“wherein said detecting the plurality of distinct multidimensional points comprises normalization of whitespace;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 6
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“wherein said detecting the plurality of distinct multidimensional points comprises decreasing a numeric precision;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“wherein said detecting the plurality of distinct multidimensional points comprises decreasing a numeric precision;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 7
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“wherein: the textual commands represented by the original plurality of multidimensional points are database statements; said accuracy of the reconstructive model comprises anomaly detection accuracy;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“wherein: the textual commands represented by the original plurality of multidimensional points are database statements; said accuracy of the reconstructive model comprises anomaly detection accuracy;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 8
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“wherein the original plurality of multidimensional points contains at least a hundred times as many multidimensional points as the plurality of distinct multidimensional points;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“wherein the original plurality of multidimensional points contains at least a hundred times as many multidimensional points as the plurality of distinct multidimensional points;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 9
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“wherein said increasing the accuracy of the reconstructive model comprises applying stochastic gradient descent to a denoising autoencoder;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“wherein said increasing the accuracy of the reconstructive model comprises applying stochastic gradient descent to a denoising autoencoder;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 10
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“the reconstructive model inferring without using a distance measurement;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“the reconstructive model inferring without using a distance measurement;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claim 11
Step 2A Prong 1:
Claim 11 recites:
“[One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors], cause: detecting a plurality of distinct multidimensional points in an original plurality of multidimensional points that contains duplicates;” Detecting a plurality of distinct multidimensional points is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
“increasing, based on duplicates in the original plurality of multidimensional points, a respective observed frequency of each distinct multidimensional point in the plurality of distinct multidimensional points;” Increasing a respective observed frequency is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
“generating, from the feature vector, a reconstruction of a distinct multidimensional point [by a reconstructive model];” Generating a reconstruction of the distinct multidimensional point is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
“increasing, based on said increasing said observed frequency of the distinct multidimensional point, a scaled error of the reconstruction of the distinct multidimensional point;” Increasing a scaled error of the reconstruction is an action that can be performed mentally with the aid of pen and paper, and is therefore a mental process.
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors;” Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)).
“encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
“by a reconstructive model;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
“training the reconstructive model, including based on the scaled error of the reconstruction of the distinct multidimensional point, increasing accuracy of the reconstructive model;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors;” Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)) which cannot provide an inventive concept.
“encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
“by a reconstructive model;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
“training the reconstructive model, including based on the scaled error of the reconstruction of the distinct multidimensional point, increasing accuracy of the reconstructive model;” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
Claims 12-20 are non-transitory computer-readable media claims that recites identical limitations to method claims 2-10. Therefore, claims 12-20 are rejected using the same rationale as claims 2-10.
Claim 21
Step 2A Prong Two
This judicial exception is not integrated into a practical application because the additional elements are as follows:
“wherein the reconstructive model is a principal component analysis (PCA);” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)).
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements are as follows:
“wherein the reconstructive model is a principal component analysis (PCA);” The limitation amounts to merely indicating a field of use or technological environment in which to apply a judicial exception. This does not amount to significantly more than the exception itself (MPEP 2106.05(h)) and cannot provide an inventive concept.
Even when considered in combination, these additional elements represent mere instructions to apply an exception and therefore do not provide an inventive concept. The claim is ineligible.
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.
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.
Claim(s) 1, 2, 11, and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Yin et al. (CN111694802A, see attached translation), hereinafter Yin, in view of Kvernvik et al. (US 20190334784 A1), hereinafter Kvernvik, Lin et al. (US 20210073644 A1), hereinafter Lin, Oishi et al. (US 20250078855 A1), hereinafter Oishi, de Zambotti et al. (US 20240016456 A1), hereinafter Zambotti.
Regarding claim 1, Yin teaches,
A method comprising: detecting a plurality of distinct multidimensional points in an original plurality of multidimensional points that contains duplicates; increasing, based on duplicates in the original plurality of multidimensional points, a respective observed frequency (Para 0010, count the number of times each feature value appears in the n feature values in the deduplicated dataset) of each distinct multidimensional point in the plurality of distinct multidimensional points [Para 0009, The initial dataset is sampled to obtain a sampled dataset. The initial dataset includes N feature values belonging to the same attribute. The sampled dataset includes n feature values from the N feature values, where n is an integer less than N; Para 0010, Perform a deduplication operation on the n feature values to obtain a deduplicated dataset, and count the number of times each feature value appears in the n feature values in the deduplicated dataset];
wherein the method is performed by one or more computers [Paras 0046-0049, this application provides an electronic device, comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the steps of the deduplication information acquisition method provided in this application].
Yin does not teach encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points; generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model; increasing, based on increasing frequency of distinct point, a scaled error of reconstruction of the distinct point; increasing, based on the scaled error of the reconstruction of the distinct point, accuracy of the reconstructive model.
Kvernvik teaches,
Increasing (Para 0085, A larger reconstruction error), based on increasing frequency of distinct point (Para 0087, anomalous behaviour is reflected in the observed parameter values), a scaled error of reconstruction of the distinct point [Para 0085, A larger reconstruction error will thus be used as an indication of anomalous behaviour reflected by the observed parameter values under consideration; Para 0087, If the combined reconstruction error or anomaly metric is outside the normal profile, then an anomalous behaviour is reflected in the observed parameter values for the time window].
Kvernvik is analogous to the claimed invention as they both relate to reconstructive models. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Kvernvik and provide increasing a scaled error of reconstruction of the distinct point based on increasing frequency of distinct point in order to improve anomaly detection analyses.
Yin-Kvernvik teach the above limitations of claim 1 including the scaled error of the reconstruction of the distinct point (Kvernvik, para 0085).
Yin-Kvernvik do not teach encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points; generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model; increasing accuracy of the reconstructive model.
Lin teaches,
training the reconstructive model [Para 0035, After the derivative is computed, a weight update can be performed by updating all the weights of the filters. For instance, the weights can be updated so that they change in the opposite direction of the gradient], increasing accuracy of reconstructive model (Para 0084, in order to minimize (or optimize) the loss function) [Para 0060, The candidate filters can be determined and removed from the complex layers or branches of the trained neural network 302 in a way that preserves the local features of the trained neural network 302, ensuring that the compressed versions of the complex layers or branches output features that are similar to the features output from the original (uncompressed) layers or branches. Using a layer including a set of filters (over one or multiple channels) as an example, the local features of the set of filters can be preserved by generating a copy of the set of filters (referred to as a duplicate set of filters), and using the original set of filters as supervision (e.g., by keeping the original set of filters fixed) as the duplicate set of filters are updated and as one or more candidate filters are removed from the duplicate set of filters. For example, as described below, the loss minimization engine 312 can minimize a loss function (e.g., including an error function, such as mean squared error, and the penalty applied by the penalty engine 314) to maintain the outputs of the original set of filters as weights, scaling factors, and/or other parameters of the duplicate filters are updated and as the candidate filters or parameters (e.g., weights) of the candidate filters are removed from the duplicate set of filters; Para 0083, At block 610, the process 600 includes minimizing a loss function of an error between the output of the set of filters and the output of the duplicate set of filters and a penalty applied to one or more scaling factors associated with the duplicate set of filters; Para 0084, the process 600 can include minimizing the loss function by iteratively determining the error between the output of the set of filters and the output of the duplicate set of filters and the penalty applied to the one or more scaling factors associated with the duplicate set of filters… as described above with reference to Equation (3), various iterations of the loss function can be performed in order to minimize (or optimize) the loss function.].
Lin is analogous to the claimed invention as they both relate to data deduplication. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Lin and provide a scaled error in order to tune model weights for improved accuracy.
Yin- Kvernvik-Lin teach the above limitations of claim 1 including the plurality of distinct multidimensional points (Yin, para 0010).
Yin- Kvernvik-Lin do not teach encoding, into a feature vector, a multidimensional point; generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model.
Oishi teaches,
encoding, into a feature vector, a multidimensional point [Para 0078, As described above, the information generation unit 12b generates a concept embedding vector (feature information) of the concept specification signal (related information) of the target voice signal (target signal); Para 0079, The information generation unit 12b encodes the concept specification signal (related information) into an image feature vector (first multidimensional vector). The information generation unit 12b encodes the mixed voice signal (mixed signal) into a voice feature vector (second multidimensional vector). The information generation unit 12b derives a similarity profile (chronological similarity) between the image feature vector and the voice feature vector. The information generation unit 12b generates a result of the weighted sum of the similarity profile and the mixed voice signal (mixed signal) as a concept embedding vector].
Oishi is analogous to the claimed invention as they both relate to feature extraction. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Oishi and provide encoding a multidimensional point into a feature vector in order to [Oishi, para 0080] improve the accuracy of feature extraction from mixed signals.
Yin- Kvernvik-Lin-Oishi do not teach generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model.
Zambotti teaches,
generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model [Para 0110, The multi-feature input 552 can include the previously described set of features, such as the multi-rate features; Para 0111, The decoder 555 uses the hidden information to reconstruct the original input features, as shown at 556. The hidden information can include the pseudo-features and the combinatorial functions used to compress or generate the pseudo-features. With typical multidimensional data, the autoencoder 550 can reconstruct the input features with low reconstruction error, at shown at 558].
Zambotti is analogous to the claimed invention as they both relate to processing multidimensional data. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Zambotti and provide reconstructing a multidimensional point from a feature vector in order to [Zambotti, para 0060] improve ML systems by identifying errors.
Regarding claim 2, Yin-Kvernvik-Lin-Oishi-Zambotti teach the limitations of claim 1.
Lin further teaches,
training reconstructive model with a training corpus that consists of plurality of distinct multidimensional points [Para 0060, duplicate filters are updated and as the candidate filters or parameters (e.g., weights) of the candidate filters are removed from the duplicate set of filters; Para 0084, the process 600 can include minimizing the loss function by iteratively determining the error between the output of the set of filters and the output of the duplicate set of filters and the penalty applied to the one or more scaling factors associated with the duplicate set of filters… as described above with reference to Equation (3), various iterations of the loss function can be performed in order to minimize (or optimize) the loss function].
Lin is analogous to the claimed invention as they both relate to data deduplication. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Lin and provide training reconstructive model [Lin, para 0032] in order to tune the model’s accuracy.
Regarding claim 11, Yin teaches,
One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors [Para 0050, this application provides a non-transitory computer-readable storage medium storing computer instructions, the computer instructions being used to cause the computer to execute the steps of the deduplication information acquisition method provided in this application; Para 0128, A processor can process instructions that execute within an electronic device, including instructions stored in or on memory], cause: detecting a plurality of distinct multidimensional points in an original plurality of multidimensional points that contains duplicates; increasing, based on duplicates in the original plurality of multidimensional points, a respective observed frequency (Para 0010, count the number of times each feature value appears in the n feature values in the deduplicated dataset) of each distinct multidimensional point in the plurality of distinct multidimensional points [Para 0009, The initial dataset is sampled to obtain a sampled dataset. The initial dataset includes N feature values belonging to the same attribute. The sampled dataset includes n feature values from the N feature values, where n is an integer less than N; Para 0010, Perform a deduplication operation on the n feature values to obtain a deduplicated dataset, and count the number of times each feature value appears in the n feature values in the deduplicated dataset].
Yin does not teach encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points; generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model; increasing, based on increasing frequency of distinct point, a scaled error of reconstruction of the distinct point; increasing, based on the scaled error of the reconstruction of the distinct point, accuracy of the reconstructive model.
Kvernvik teaches,
Increasing (Para 0085, A larger reconstruction error), based on increasing frequency of distinct point (Para 0087, anomalous behaviour is reflected in the observed parameter values), a scaled error of reconstruction of the distinct point [Para 0085, A larger reconstruction error will thus be used as an indication of anomalous behaviour reflected by the observed parameter values under consideration; Para 0087, If the combined reconstruction error or anomaly metric is outside the normal profile, then an anomalous behaviour is reflected in the observed parameter values for the time window].
Kvernvik is analogous to the claimed invention as they both relate to reconstructive models. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Kvernvik and provide increasing a scaled error of reconstruction of the distinct point based on increasing frequency of distinct point in order to improve anomaly detection analyses.
Yin-Kvernvik teach the above limitations of claim 1 including the scaled error of the reconstruction of the distinct point (Kvernvik, para 0085).
Yin-Kvernvik do not teach encoding, into a feature vector, a multidimensional point from the plurality of distinct multidimensional points; generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model; increasing accuracy of the reconstructive model.
Lin teaches,
training the reconstructive model [Para 0035, After the derivative is computed, a weight update can be performed by updating all the weights of the filters. For instance, the weights can be updated so that they change in the opposite direction of the gradient], increasing accuracy of reconstructive model (Para 0084, in order to minimize (or optimize) the loss function) [Para 0060, The candidate filters can be determined and removed from the complex layers or branches of the trained neural network 302 in a way that preserves the local features of the trained neural network 302, ensuring that the compressed versions of the complex layers or branches output features that are similar to the features output from the original (uncompressed) layers or branches. Using a layer including a set of filters (over one or multiple channels) as an example, the local features of the set of filters can be preserved by generating a copy of the set of filters (referred to as a duplicate set of filters), and using the original set of filters as supervision (e.g., by keeping the original set of filters fixed) as the duplicate set of filters are updated and as one or more candidate filters are removed from the duplicate set of filters. For example, as described below, the loss minimization engine 312 can minimize a loss function (e.g., including an error function, such as mean squared error, and the penalty applied by the penalty engine 314) to maintain the outputs of the original set of filters as weights, scaling factors, and/or other parameters of the duplicate filters are updated and as the candidate filters or parameters (e.g., weights) of the candidate filters are removed from the duplicate set of filters; Para 0083, At block 610, the process 600 includes minimizing a loss function of an error between the output of the set of filters and the output of the duplicate set of filters and a penalty applied to one or more scaling factors associated with the duplicate set of filters; Para 0084, the process 600 can include minimizing the loss function by iteratively determining the error between the output of the set of filters and the output of the duplicate set of filters and the penalty applied to the one or more scaling factors associated with the duplicate set of filters… as described above with reference to Equation (3), various iterations of the loss function can be performed in order to minimize (or optimize) the loss function.].
Lin is analogous to the claimed invention as they both relate to data deduplication. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Lin and provide a scaled error in order to tune model weights for improved accuracy.
Yin-Kvernvik-Lin teach the above limitations of claim 1 including the plurality of distinct multidimensional points (Yin, para 0010).
Yin-Kvernvik-Lin do not teach encoding, into a feature vector, a multidimensional point; generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model.
Oishi teaches,
encoding, into a feature vector, a multidimensional point [Para 0078, As described above, the information generation unit 12b generates a concept embedding vector (feature information) of the concept specification signal (related information) of the target voice signal (target signal); Para 0079, The information generation unit 12b encodes the concept specification signal (related information) into an image feature vector (first multidimensional vector). The information generation unit 12b encodes the mixed voice signal (mixed signal) into a voice feature vector (second multidimensional vector). The information generation unit 12b derives a similarity profile (chronological similarity) between the image feature vector and the voice feature vector. The information generation unit 12b generates a result of the weighted sum of the similarity profile and the mixed voice signal (mixed signal) as a concept embedding vector].
Oishi is analogous to the claimed invention as they both relate to feature extraction. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Oishi and provide encoding a multidimensional point into a feature vector in order to [Oishi, para 0080] improve the accuracy of feature extraction from mixed signals.
Yin-Kvernvik-Lin-Oishi do not teach generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model.
Zambotti teaches,
generating, from feature vector, a reconstruction of a distinct multidimensional point by a reconstructive model [Para 0110, The multi-feature input 552 can include the previously described set of features, such as the multi-rate features; Para 0111, The decoder 555 uses the hidden information to reconstruct the original input features, as shown at 556. The hidden information can include the pseudo-features and the combinatorial functions used to compress or generate the pseudo-features. With typical multidimensional data, the autoencoder 550 can reconstruct the input features with low reconstruction error, at shown at 558].
Zambotti is analogous to the claimed invention as they both relate to processing multidimensional data. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Zambotti and provide reconstructing a multidimensional point from a feature vector in order to [Zambotti, para 0060] improve ML systems by identifying errors.
Claim 12 is a non-transitory computer-readable media claim that recites identical limitations to method claim 2. Therefore, claim 12 is rejected using the same rationale as claim 2.
Claim(s) 3 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, and Zambotti, and in further view of Marti et al. (US 20130170762 A1), hereinafter Marti.
Regarding claim 3, Yin-Kvernvik-Lin-Oishi-Zambotti teach the limitations of claim 1 including multidimensional points (Yin, para 0009).
Yin-Kvernvik-Lin-Oishi-Zambotii do not teach generating a batch that represents more points than the batch contains.
Marti teaches,
generating a batch (Para 0073, improved reference data) that represents more points (Para 0073, reference data sets 64 can be increased) than the batch contains (Para 0073, batch 15 may be added to the reference data sets 64) [Para 0073, the data sets 1-11 in the batch 15 may be added to the reference data sets 64. In this way, the number of reference data sets 64 can be increased to give improved reference data for future correlation analysis processing].
Marti is analogous to the claimed invention as they both relate to data compression. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Marti and provide generating a batch with more data than the original batch contained [Marti, para 0073] in order to give improved reference data for future analysis processing.
Claim 13 is a non-transitory computer-readable media claim that recites identical limitations to method claim 3. Therefore, claim 13 is rejected using the same rationale as claim 3.
Claim(s) 4 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, and Zambotti, and in further view of Zhang et al. (CN113408301A, see attached translation), hereinafter Zhang.
Regarding claim 4, Yin-Kvernvik-Lin-Oishi-Zambotti teach the limitations of claim 1 including multidimensional points and the original plurality of multidimensional points (Yin, para 0009).
Yin-Kvernvik-Lin-Oishi-Zambotti do not teach wherein each point in plurality of points represents a respective textual command.
Zhang teaches,
wherein each point in plurality of points represents a respective textual command [Para 0022, the query text in the initial training samples of the preset text matching model, i.e. the keywords input into the preset text matching model, is clustered. Then, the clustered query text is deduplicated and corrected according to the category and the corresponding sample timestamp. That is, multiple initial training samples generated within a certain period of time are deduplicated or the labels corresponding to the samples are corrected, and finally, target model training samples with higher sample data quality are obtained].
Zhang is analogous to the claimed invention as they both relate to data duplication. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Zhang and provide data points as textual commands [Zhang, para 0022] in order to obtain higher quality sample data in particular machine learning models such as text matching models.
Claim 14 is a non-transitory computer-readable media claim that recites identical limitations to method claim 4. Therefore, claim 14 is rejected using the same rationale as claim 4.
Claim(s) 5 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, Zambotti, and Zhang, and in further view of Shin et al. (US 20220171769 A1), hereinafter Shin.
Regarding claim 5, Yin-Kvernvik-Lin-Oshi-Zambotti-Zhang teach the limitations of claim 4 including said detecting the plurality of distinct multidimensional points (claim 1: Yin, paras 0009 and 0010).
Yin-Kvernvik-Lin-Oishi-Zambotti-Zhang do not teach normalization of whitespace.
Shin teaches,
normalization of whitespace [Para 0029, Sanitization operation 201 may include setting all letters to one case (e.g., replacing uppercase characters with lowercase characters, or replacing all lowercase characters with uppercase characters), eliminating non-alphabetic characters (e.g., punctuation, numbers, or the like), normalizing white space (e.g., replacing multiple spaces between words, such as the triple space between “the” and “sentence” in search data 103, with a single space), and/or other suitable operations].
Shin is analogous to the claimed invention as they both relate to textual data manipulation. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Shin and provide normalization of whitespace in order to improve NPL search queries and matching.
Claim 15 is a non-transitory computer-readable media claim that recites identical limitations to method claim 5. Therefore, claim 15 is rejected using the same rationale as claim 5.
Claim(s) 6 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, Zambotti, and Zhang, and in further view of O'Hare et al. (US 20190340262 A1), hereinafter O’Hare.
Regarding claim 6, Yin-Kvernvik-Lin-Oishi-Zambotti-Zhang teach the limitations of claim 4 including said detecting the plurality of distinct multidimensional points (claim 1: Yin, paras 0009 and 0010).
Yin-Kvernvik-Lin-Oishi-Zambotti-Zhang do not teach decreasing a numeric precision.
O’Hare teaches,
decreasing a numeric precision [Para 0050, As part of the deduplication process, embodiments will iteratively truncate 414 the hash values associated with these unwritten data chunks in order to determine 416 if data chunk C D E F has already been sequentially stored in physical storage in no more than two locations. This determination 416 is made by comparing each iterative hash string 524-528 to the values stored in hash table 520 until a sequential match is found between a first and last hash value in no more than two physical locations].
O’Hare is analogous to the claimed invention as they both relate to data deduplication. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of O’Hare and provide decreasing a numeric precision in order to [O’Hare, Abstract] reduce computational speeds and storage requirements.
Claim 16 is a non-transitory computer-readable media claim that recites identical limitations to method claim 6. Therefore, claim 16 is rejected using the same rationale as claim 6.
Claim(s) 7 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, Zambotti, and Zhang, and in further view of Kascenas et al. (Denoising Autoencoders for Unsupervised Anomaly Detection in Brain MRI, published 2022), hereinafter Kascenas.
Regarding claim 7, Yin-Kvernvik-Lin-Oishi-Zambotti-Zhang teach the limitations of claim 4 including the original plurality of multidimensional points (claim 1: Yin, paras 0009 and 0010) and said accuracy of the reconstructive model (claim 1, Lin, para 0084).
Zhang further teaches,
the textual commands are database statements [Para 0032, Suppose a user enters a query text into a pre-defined text matching model such as a knowledge question-answering system, and the system returns 20 relevant texts.].
Zhang is analogous to the claimed invention as they both relate to data duplication. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Zhang and provide data points as textual commands are database statements [Zhang, para 0022] in order to obtain higher quality sample data in particular machine learning models such as text matching models.
Yin-Kvernvik-Lin-Oishi-Zambotti-Zhang teach do not teach anomaly detection accuracy.
Kascenas teaches,
anomaly detection accuracy [Figure 1, Our denoising autoencoder anomaly detection method. During training (top), noise is added to the foreground of the healthy image, and the network is trained to reconstruct the original image. At test time (bottom), the pixelwise post processed reconstruction error is used as the anomaly score].
Kascenas is analogous to the claimed invention as they both relate to autoencoders. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Kascenas and provide anomaly detection accuracy [Kascenas, Sect 1, para 4] in order to improve intensity threshold for machine learning systems.
Claim 17 is a non-transitory computer-readable media claim that recites identical limitations to method claim 7. Therefore, claim 17 is rejected using the same rationale as claim 7.
Claim(s) 8 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, and Zambotti, and in further view of Zuluaga et al. (US 20190156442 A1), hereinafter Zuluaga.
Regarding claim 8, Yin-Kvernvik-Lin-Oishi-Zambotti teach the limitations of claim 1 including the original plurality of multidimensional points and the plurality of distinct multidimensional points (Yin, para 0032).
Yin-Kvernvik-Lin-Oishi-Zambotti do not teach wherein original points contains at least a hundred times as many points as plurality of distinct points.
Zuluaga teaches,
wherein original points contains at least a hundred times as many points as plurality of distinct points [Para 0024, Data deduplication is a data compression process which may match duplicate copies of repeated data such as duplicate web listings. In the deduplication process, web listings may be processed to identify web listings that are a match to one another. Often a stored web listing or master copy is compared to a newly received web listing. When a match occurs, the redundant web listing may be replaced with a small reference (e.g., bit value, pointer, URL, etc.) that points to the web listing, rather than storing a duplicate copy of the web listing and its images, description, reviews, etc. within a storage inventory (e.g., a file, a table, a data store, a database file, etc.) Because the same web listing may occur dozens or even hundreds of time, the amount of data that is stored and maintained may be greatly reduced by deduplication. When a subsequent search is performed, only a single web listing may be provided which is used to represent a group of duplicate web listings which can be found across multiple sites].
Zuluaga is analogous to the claimed invention as they both relate to deduplication. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Zuluaga and provide original points containing at least a hundred times as many points as plurality of distinct points in order to provide a comprehensive training set to improve machine learning models.
Claim 18 is a non-transitory computer-readable media claim that recites identical limitations to method claim 8. Therefore, claim 18 is rejected using the same rationale as claim 8.
Claim(s) 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, and Zambotti, and in further view of Liang et al. (Training Stacked Denoising Autoencoders for Representation Learning, published 2021), hereinafter Liang.
Regarding claim 9, Yin-Kvernvik-Lin-Oishi-Zambotti teach the limitations of claim 1 including said increasing the accuracy of the reconstructive model (Lin, para 0084).
Yin-Kvernvik-Lin-Oishi-Zambotti do not teach applying stochastic gradient descent to a denoising autoencoder.
Liang teaches,
applying stochastic gradient descent to a denoising autoencoder [Abstract, We implement stacked denoising autoencoders, a class of neural networks that are capable of learning powerful representations of high dimensional data. We describe stochastic gradient descent for unsupervised training of autoencoders, as well as a novel genetic algorithm based approach that makes use of gradient information].
Liang is analogous to the claimed invention as they both relate to autoencoders. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Liang and provide anomaly applying stochastic gradient descent to a denoising autoencoder in order to enhance machine learning systems by implementing a processing methodology that improves efficiency and scalability with frequent updates.
Claim 19 is a non-transitory computer-readable media claim that recites identical limitations to method claim 9. Therefore, claim 19 is rejected using the same rationale as claim 9.
Claim(s) 10 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, and Zambotti, and in further view of Ren et al. (US 20230196597 A1), hereinafter Ren.
Regarding claim 10, Yin-Kvernvik-Lin-Oishi-Zambotti teach the limitations of claim 1 including the reconstructive model (Yin, para 0029).
Yin-Kvernvik-Lin-Oishi-Zambotti do not teach model inferring without using a distance measurement.
Ren teaches,
model inferring without using a distance measurement [Para 0076, At step 808, the system may generate a first three-dimensional model of the environment. In some embodiments, the three-dimensional model may be a three-dimensional point cloud, for example. The system may generate the first three-dimensional model based on the image sequence alone, i.e. without the set of distance measurements].
Ren is analogous to the claimed invention as they both relate to autoencoders. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Ren and provide model inferring without using a distance measurement in order to improve model robustness when handling outliers.
Claim 20 is a non-transitory computer-readable media claim that recites identical limitations to method claim 10. Therefore, claim 20 is rejected using the same rationale as claim 10.
Claim(s) 21 is rejected under 35 U.S.C. 103 as being unpatentable over Yin in view of Kvernvik, Lin, Oishi, and Zambotti, and in further view of Consoli et al. (US 20200219627 A1), hereinafter Consoli.
Regarding claim 21, Yin-Kvernvik-Lin-Oishi-Zambotti teach the limitations of claim 1 including the reconstructive model (Lin, Para 0035).
Yin-Kvernvik-Lin-Oishi-Zambotti do not teach wherein model is a principal component analysis (PCA).
Consoli teaches,
wherein model is a principal component analysis (PCA) [Abstract, A method of clustering or grouping subjects that are similar to one another. A dataset contains, for each subject, a set of quantitative values which each represent a respective clinical or pathological feature of that subject. A principle component analysis, PCA, is performed on the dataset].
Consoli is analogous to the claimed invention as they both relate to utilizing PCA. Therefore, it would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yin’s teachings to incorporate the teachings of Consoli and provide principle component analysis in order to [Consoli, para 0008] improve diagnosis capabilities, selection of appropriate treatment options and prediction of probable subject outcomes.
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 SYED RAYHAN AHMED whose telephone number is (571)270-0286. The examiner can normally be reached Mon-Fri ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, David Yi can be reached at (571) 270-7519. 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.
/SYED RAYHAN AHMED/Examiner, Art Unit 2126
/DAVID YI/Supervisory Patent Examiner, Art Unit 2126
Prosecution Insights