DETAILED ACTION
This action is responsive to the Application filed on 10/16/2023. Claims 1-20 are pending in the case. Claims 1, 8, and 15 are independent claims.
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 .
Information Disclosure Statement
The listing of references in the specification is not a proper information disclosure statement. 37 CFR 1.98(b) requires a list of all patents, publications, or other information submitted for consideration by the Office, and MPEP § 609.04(a) states, "the list may not be incorporated into the specification but must be submitted in a separate paper." Therefore, unless the references have been cited by the examiner on form PTO-892, they have not been considered.
Drawings
Figure 1 should be designated by a legend such as --Prior Art-- because only that which is old is illustrated. See MPEP § 608.02(g). Corrected drawings in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. The replacement sheet(s) should be labeled “Replacement Sheet” in the page header (as per 37 CFR 1.84(c)) so as not to obstruct any portion of the drawing figures. If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
The drawings are objected to because the "Unlabeled data clusters" component seen in figures 2A, 3A, 4A, and 5A should be provided with a reference number given the repeated appearance of the same component in multiple embodiments. Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. The figure or figure number of an amended drawing should not be labeled as “amended.” If a drawing figure is to be canceled, the appropriate figure must be removed from the replacement sheet, and where necessary, the remaining figures must be renumbered and appropriate changes made to the brief description of the several views of the drawings for consistency. Additional replacement sheets may be necessary to show the renumbering of the remaining figures. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Specification
The abstract of the disclosure is objected to because the sentence starting in line 3 and ending in line 6 is a run-on sentence and difficult to interpret. The examiner recommends that the applicant formats the sentence in a more clear and concise manner, potentially dividing it into multiple sentences. A corrected abstract of the disclosure is required and must be presented on a separate sheet, apart from any other text. See MPEP § 608.01(b).
The disclosure is objected to because of the following informalities:
Paragraph [0002], line 6, “…maximize the monetization online platforms…” should read “…maximize the monetization of online platforms…”
Paragraph [0004], line 4, “having some slightly changes features” should read “having some slightly changed features”
Paragraph [0026], line 5, “…and/or system” should read “and/or systems…”
In the following locations the ‘weakly labeled training data’ is referred to using ref no. 260 while the drawings label the ‘weakly labeled training data’ as ref no. 270, corrections should be made
Paragraph 35, line 7
Paragraph 36, line 2
Paragraph 36, line 3
Paragraph 36, line 16
In the following locations ref no. 270, introduced as ‘weakly labeled training data,’ is modified to be ‘weakly labeled training data storage.’ These descriptions are not equivalent and corrections should be made
Paragraph 38, lines 8 and 13
Paragraph 39, lines 13 and 16
Paragraph 48, line 13
Paragraph 47, line 7, ‘self-training generator’ is referred to using ref no. 450 while the drawings label ‘self-training generator’ as ref no. 330, corrections should be made
In paragraph [0043], the ‘weakly labeled training data’ (ref no. 270) is further defined as Daug, but some later references to this ‘weakly labeled training data’ use ref no. 270 and/or Daug, in all instances both labels should be provided with an ‘or’ to maintain consistency
Paragraph [0051], line 3, ‘generative augmentation generator’ is referred to using ref no. 3430, this should be corrected to ref no. 340
Appropriate correction is required.
Claim Objections
Claims 2-6, 9-13, and 15-20 are objected to because of the following informalities:
Claim 2, line 1, recites “…wherein the generating the weakly…” should read “…wherein generating the weakly….”
Claim 3, line 8, recites “…the cluster label if one of the…” should read “…the cluster label is one of the…”
Claim 9, line 1, recites “…wherein the generating the weakly…” should read “…wherein generating the weakly….”
Claim 10, line 1, “The method of claim 9…” should read “The medium of claim 9…”
Claim 10, line 8 , recites “…the cluster label if one of the…” should read “…the cluster label is one of the…”
Claim 16, line 1, recites “…wherein the generating the weakly…” should read “…wherein generating the weakly….”
Claim 17, line 8, recites “…the cluster label if one of the…” should read “…the cluster label is one of the…”
Claims 5-6 and claims 12-13 should be combined into singular claims following that of claim 19 in order to remain consistent throughout the entire list of claims
Claim 15, lines 12-13, recite “an augmented data-based model training engine implemented by a processor and configured.” Note that you can’t just say something is “configured.” Define what the model is configured to do or, if it is meant to be configured to do the steps following it in the claim, add a “for” to the end of the sentence
In claims 3, 10, and 17, the generating step should be at the same indentation level as the previous steps
Claims 4, 11, and 18-20 inherit the objections from the claims upon which they depend
Appropriate correction is required.
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 towards an abstract idea without significantly more.
Step 1: Claims 1-7 are directed towards a method, Claims 8-14 are directed towards an article of manufacture, and Claims 15-20 are directed towards a machine/system. Therefore, Claims 1-20 are directed towards one of the 4 statutory categories; process, machine, manufacture or composition of matter.
With respect to claim 1:
Step 2A Prong 1: The claim is directed to a judicial exception.
generating weakly labeled training data based on the supervised training data and the unlabeled data clusters, wherein the weakly labeled training data includes data samples each of which is assigned with one of the plurality of labels via consistent self-training so that a data sample in the supervised training data with a label and a data sample from the weakly labeled training data with the same label have varying characteristics (Mental Process: One could generate weakly labeled training data based on supervised training data and unlabeled data clusters mentally or using a pen and paper. The mention of consistent self-training does not imply that the generating has to include that process, just that the final data group has data labeled in that way)
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application.
Additional Elements:
receiving supervised training data and unlabeled data clusters, wherein the supervised training data include data samples each of which has a label from a plurality of labels and each of the unlabeled data clusters includes multiple unlabeled data samples with varying features (Amounts to necessary data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
obtaining augmented training data based on the supervised training data and the weakly labeled training data (Mere data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
Training, via machine learning, a robust content categorization model based on the augmented training data (High level machine learning. Any model could be trained to be a robust content categorization model. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Re-evaluation of Insignificant Extra-Solution Activities:
receiving supervised training data and unlabeled data clusters, wherein the supervised training data include data samples each of which has a label from a plurality of labels and each of the unlabeled data clusters includes multiple unlabeled data samples with varying features (“Receiving or transmitting data over a network” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II)
obtaining augmented training data based on the supervised training data and the weakly labeled training data (“Storing and retrieving information in memory” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II)
Additional Elements:
Training, via machine learning, a robust content categorization model based on the augmented training data (High level machine learning. Any model could be trained to be a robust content categorization model. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
With respect to claim 2:
Step 2A Prong 1: The claim is directed to a judicial exception, including those inherited from claim 1 via dependency.
creating the weakly labeled training data based on the unlabeled data clusters and the base pseudo label prediction model (Mental Process: One could mentally create the weakly labeled training data based on the unlabeled data and base pseudo label prediction model using pen and paper. Note that the use of the model is an “apply it” as seen below.)
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application.
Additional Elements:
The method of claim 1 (See above)
training, via machine learning, a base pseudo label prediction model based on the supervised training data (High level machine learning. Any model could be trained to be a base pseudo label predication model. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
creating the weakly labeled training data based on the unlabeled data clusters and the base pseudo label prediction model (High level machine learning. Simply using a generic model trained using ‘machine learning’ to perform a task. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Additional Elements:
The method of claim 1 (See above)
training, via machine learning, a base pseudo label prediction model based on the supervised training data (High level machine learning. Any model could be trained to be a base pseudo label predication model. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
creating the weakly labeled training data based on the unlabeled data clusters and the base pseudo label prediction model (High level machine learning. Simply using a generic model trained using ‘machine learning’ to perform a task. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
With respect to claim 3:
Step 2A Prong 1: The claim is directed to a judicial exception, including those inherited from claim 2 via dependency.
predicting, for each unlabeled data sample in the unlabeled data cluster, a pseudo label based on the base pseudo label prediction model, wherein the predicted pseudo label is one of the plurality of labels (Mental Process: One could predict a label for each sample from a plurality of labels based on a model using evaluation and pen and paper. Not that use of the model is an “apply it” as seen below)
determining a cluster label for the unlabeled data cluster based on the pseudo labels predicted respectively for the unlabeled data samples in the unlabeled data cluster, wherein the cluster label if one of the plurality of labels (Mental Process: One could determine the most suitable label for the data cluster using a pen and paper and mere observation and evaluation of the predicted labels)
assigning the cluster label to each of the unlabeled data samples in the unlabeled data cluster (Mental Process: One could assign the selected cluster label to each of the data samples mentally, or using pen and paper)
generating, based on the data samples in the unlabeled data clusters with newly assigned cluster labels, the weakly labeled training data (Mental Process: Given the data samples in the unlabeled data clusters with newly assigned labels, one could mentally, or using pen and paper, generate a group of data, e.g. the weakly labeled training data)
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application.
Additional Elements:
The method of claim 2 (See above)
predicting, for each unlabeled data sample in the unlabeled data cluster, a pseudo label based on the base pseudo label prediction model, wherein the predicted pseudo label is one of the plurality of labels (High level machine learning. Simply using a generic model trained using machine learning to perform a task. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Additional Elements:
The method of claim 2 (See above)
predicting, for each unlabeled data sample in the unlabeled data cluster, a pseudo label based on the base pseudo label prediction model, wherein the predicted pseudo label is one of the plurality of labels (High level machine learning. Simply using a generic model trained using machine learning to perform a task. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
With respect to claim 4:
Step 2A Prong 1: The claim is directed to a judicial exception, including those inherited from claim 3 via dependency.
processing the pseudo labels predicted for the unlabeled data samples in the unlabeled data cluster (Mental Process: Processing is a generic term for using observation or evaluation to understand something, which one could do mentally)
selecting one of the pseudo labels as the cluster label according to a predetermined function based on the at least one metric (Mental Process: One could perform selection of a pseudo label according to a predetermined function based on a metric, using pen and paper or mentally)
Step 2A Prong 2: The judicial exceptions as a whole are not integrated into a practical application.
Additional Elements:
The method of claim 3 (See above)
obtaining at least one metric associated with the pseudo labels for data samples in the unlabeled data cluster (Mere data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exceptions.
Re-evaluation of Insignificant Extra-Solution Activities:
obtaining at least one metric associated with the pseudo labels for data samples in the unlabeled data cluster (“Storing and retrieving information in memory” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II)
Additional Elements:
The method of claim 3 (See above)
With respect to claim 5:
Step 2A Prong 1: The claim is directed to a judicial exception, including those inherited from claim 4 via dependency.
Step 2A Prong 2: The judicial exceptions as a whole are not integrated into a practical application.
Additional Elements:
The method of claim 4… (See above)
the at least one metric includes a confidence score for each of the pseudo labels, or a frequency for each of the pseudo labels (Describes the information that the abstract idea operates on rather than an additional element to integrate the exception into a practical application, see MPEP § 2106.05(e), which discusses other meaningful limitations)
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Additional Elements:
The method of claim 4… (See above)
the at least one metric includes a confidence score for each of the pseudo labels, or a frequency for each of the pseudo labels (Describes the information that the abstract idea operates on rather than an additional element to integrate the exception into a practical application, see MPEP § 2106.05(e), which discusses other meaningful limitations)
With regards to claim 6:
Step 2A Prong 1: The claim is directed to a judicial exception, including those inherited from claim 4 via dependency.
the predetermined function is used for identifying the cluster label based on one of the pseudo labels that has a maximum confidence score, or is associated with a highest frequency. (Mental Process and Mathematical Concept: One could use a predetermined function to identify the cluster label based on the pseudo label with the maximum confidence score or highest frequency using a pen and paper or mentally)
Step 2A Prong 2: The judicial exceptions as a whole are not integrated into a practical application.
Additional Elements:
The method of claim 4… (See above)
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Additional Elements:
The method of claim 4… (See above)
With respect to claim 7:
Step 2A Prong 1: The claim is directed to a judicial exception, including those inherited from claim 1 via dependency.
classifying the content based on the robust content categorization model trained based on the augmented training data including both the supervised training data and the weakly labeled training data (Mental Process: One could mentally classify content based on how a model does it. Note that use of the model is an “apply it” as seen below)
Step 2A Prong 2: The judicial exceptions as a whole are not integrated into a practical application.
Additional Elements:
The method of claim 1… (See above)
receiving content to be categorized (Amounts to necessary data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
classifying the content based on the robust content categorization model trained based on the augmented training data including both the supervised training data and the weakly labeled training data (High level machine learning. Simply using a generic model trained via ‘machine learning’ to do a task. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Re-evaluation of Insignificant Extra-Solution Activities:
receiving content to be categorized (“Receiving or transmitting data over a network” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II))
Additional Elements:
The method of claim 1… (See above)
classifying the content based on the robust content categorization model trained based on the augmented training data including both the supervised training data and the weakly labeled training data (High level machine learning. Simply using a generic model trained via ‘machine learning’ to do a task. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
With respect to claim 8:
Step 2A Prong 1: The claim is directed to a judicial exception.
generating weakly labeled training data based on the supervised training data and the unlabeled data clusters, wherein the weakly labeled training data includes data samples each of which is assigned with one of the plurality of labels via consistent self-training so that a data sample in the supervised training data with a label and a data sample from the weakly labeled training data with the same label have varying characteristics (Mental Process: One could generate weakly labeled training data based on supervised training data and unlabeled data clusters mentally or using a pen and paper. The mention of consistent self-training does not imply that the generating has to include that process, just that the final data group has data labeled in that way)
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application.
Additional Elements:
A machine-readable and non-transitory medium having information recorded thereon, wherein the information, when read by the machine, causes the machine to perform the following steps (Non-transitory machine-readable mediums are generic computer components that all store information which can be read by a generic machine to, potentially, perform a set of steps. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
receiving supervised training data and unlabeled data clusters, wherein the supervised training data include data samples each of which has a label from a plurality of labels and each of the unlabeled data clusters includes multiple unlabeled data samples with varying features (Amounts to necessary data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
obtaining augmented training data based on the supervised training data and the weakly labeled training data (Mere data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
Training, via machine learning, a robust content categorization model based on the augmented training data (High level machine learning. Any model could be trained to be a robust content categorization model. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Re-evaluation of Insignificant Extra-Solution Activities:
receiving supervised training data and unlabeled data clusters, wherein the supervised training data include data samples each of which has a label from a plurality of labels and each of the unlabeled data clusters includes multiple unlabeled data samples with varying features (“Receiving or transmitting data over a network” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II)
obtaining augmented training data based on the supervised training data and the weakly labeled training data (“Storing and retrieving information in memory” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II)
Additional Elements:
A machine-readable and non-transitory medium having information recorded thereon, wherein the information, when read by the machine, causes the machine to perform the following steps (Non-transitory machine-readable mediums are generic computer components that all store information which can be read by a generic machine to, potentially, perform a set of steps. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Training, via machine learning, a robust content categorization model based on the augmented training data (High level machine learning. Any model could be trained to be a robust content categorization model. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
With respect to claims 9-14:
See the rejections for claims 2-7 above. Each of claims 9-14 recite the same limitations as claims 2-7, the only difference being that claims 9-14 are directed towards the article of manufacture in claim 8 while claims 2-7 are directed towards the method in claim 1. See the rejection for claim 8 above, which incorporates the additional element of the machine-readable medium into the rejection, this carries down for all of claims 9-14.
With respect to claim 15:
Step 2A Prong 1: The claim is directed to a judicial exception.
generating weakly labeled training data based on the supervised training data and the unlabeled data clusters, wherein the weakly labeled training data includes data samples each of which is assigned with one of the plurality of labels via consistent self-training so that a data sample in the supervised training data with a label and a data sample from the weakly labeled training data with the same label have varying characteristics (Mental Process: One could generate weakly labeled training data based on supervised training data and unlabeled data clusters mentally or using a pen and paper. The mention of consistent self-training does not imply that the generating has to include that process, just that the final data group has data labeled in that way)
Step 2A Prong 2: The judicial exceptions are not integrated into a practical application.
Additional Elements:
a training data augmenter implemented by a processor and configured for… (Simply naming a computer component that is implemented by a processor, which is a generic computer component. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
receiving supervised training data and unlabeled data clusters, wherein the supervised training data include data samples each of which has a label from a plurality of labels and each of the unlabeled data clusters includes multiple unlabeled data samples with varying features (Amounts to necessary data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
an augmented data-based model training engine implemented by a processor and configured… (Simply naming a computer component that is implemented by a processor, which is a generic computer component. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
obtaining augmented training data based on the supervised training data and the weakly labeled training data (Mere data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
Training, via machine learning, a robust content categorization model based on the augmented training data (High level machine learning. Any model could be trained to be a robust content categorization model. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Re-evaluation of Insignificant Extra-Solution Activities:
receiving supervised training data and unlabeled data clusters, wherein the supervised training data include data samples each of which has a label from a plurality of labels and each of the unlabeled data clusters includes multiple unlabeled data samples with varying features (“Receiving or transmitting data over a network” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II))
obtaining augmented training data based on the supervised training data and the weakly labeled training data (“Storing and retrieving information in memory” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II))
Additional Elements:
a training data augmenter implemented by a processor and configured for… (Simply naming a computer component that is implemented by a processor, which is a generic computer component. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
an augmented data-based model training engine implemented by a processor and configured… (Simply naming a computer component that is implemented by a processor, which is a generic computer component. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Training, via machine learning, a robust content categorization model based on the augmented training data (High level machine learning. Any model could be trained to be a robust content categorization model. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
With respect to claims 16-19:
See the rejections for claims 2-6 above. Claims 16-19 recite the same limitations seen in claims 2-6, the only difference being that claims 16-19 are directed towards the system in claim 15 while claims 2-6 are directed towards the method in claim 1. Additionally, note that claim 19 recites the same limitations as claims 5 and 6, as if the two claims were combined into a singular claim, adding an “and” between the two limitations; In understanding the rejection for claim 19, read the rejections for both claims 5 and 6. See the rejection for claim 15 above, which incorporates the additional element of the system and its components into the rejection, this carries down for all of claims 16-20. Claim 20 introduces an additional computer component which is included in the rejection below.
With respect to claim 20:
Step 2A Prong 1: The claim is directed to a judicial exception.
classifying the content based on the robust content categorization model trained based on the augmented training data including both the supervised training data and the weakly labeled training data (Mental Process: One could mentally classify content based on how a model does it. Note that the use of the model is an “apply it” as seen below)
Step 2A Prong 2: The judicial exceptions as a whole are not integrated into a practical application.
Additional Elements:
The system of claim 15… (See above)
further comprising a content categorization engine implemented by a processor and configured for… (Simply naming a computer component that is implemented by a processor, which is a generic computer component. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
receiving content to be categorized (Amounts to necessary data gathering. Insignificant extra-solution activity, as discussed in MPEP § 2106.05(g))
classifying the content based on the robust content categorization model trained based on the augmented training data including both the supervised training data and the weakly labeled training data (High level machine learning. Simply using a generic model trained using ‘machine learning’ to perform a task. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Step 2B: The claim does not include additional elements that amount to significantly more than the judicial exception.
Re-evaluation of Insignificant Extra-Solution Activities:
receiving content to be categorized (“Receiving or transmitting data over a network” is a well-understood, routine, conventional activity when claimed in a merely generic manner (as it is in the present claim), as discussed in MPEP § 2106.05(d)(II))
Additional Elements:
The system of claim 15… (See above)
further comprising a content categorization engine implemented by a processor and configured for… (Simply naming a computer component that is implemented by a processor, which is a generic computer component. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
classifying the content based on the robust content categorization model trained based on the augmented training data including both the supervised training data and the weakly labeled training data (High level machine learning. Simply using a generic model trained using ‘machine learning’ to perform a task. 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 using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f))
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-20 are rejected under 35 U.S.C. 102(a)(1) and 102(a)(2) as being anticipated by Taccari (US 20230116417 A1).
Regarding claim 1, Taccari teaches a method (Fig. 1A), comprising:
receiving supervised training data (Fig. 1A, Ref No. 102, Note that Taccari refers to the ‘supervised training data’ as the ‘labeled data set’) and unlabeled data clusters (Fig. 1A, Ref No. 104, Note that Taccari refers the ‘unlabeled data clusters’ as the ‘unlabeled data set(s)’), wherein the supervised training data include data samples each of which has a label from a plurality of labels (Fig. 1A, Ref No. 101, See the labels ‘cat’ and ‘dog’) and each of the unlabeled data clusters includes multiple unlabeled data samples with varying features (Paragraph [0021], “Accordingly, the same object (e.g., the particular animal) may be represented in each image of the second set of unlabeled data 103 with different parameters or some variety in the parameters”, Fig. 1A, Ref No. 103, See the variations in the depictions of the cats at different points in time. Note that the parameters and variations in the depictions of the images are varying features.);
generating weakly labeled training data based on the supervised training data and the unlabeled data clusters (Paragraph [0046], “The mapping function and/or model may classify each unlabeled data item from the first group of unlabeled data to one of the labels specified for the set of labeled data”, Note that Taccari does not title the ‘weakly labeled training data’ but they are understood to be the same thing. Further, note that the ‘supervised training data’ is known as the ‘labeled data’ in Taccari, as seen above), wherein the weakly labeled training data includes data samples each of which is assigned with one of the plurality of labels via consistent self-training (Paragraph [0046], “Process 300 may include inputting (at 306) a partitioned first group of unlabeled data to the mapping function and/or model trained with the set of labeled data. The mapping function and/or model may classify each unlabeled data item from the first group of unlabeled data to one of the labels specified for the set of labeled data.” Note that the entire process described in Taccari amounts to consistent self-training, as will be seen in the citations below; this includes the steps of training a model based on a labeled data set, generating pseudo-labels for unlabeled data using some confidence threshold, augmenting the training set to include both labeled and pseudo-labeled data, and retraining a model on the augmented data set) so that a data sample in the supervised training data with a label and a data sample from the weakly labeled training data with the same label have varying characteristics (Paragraph [0021], “Accordingly, the same object [given same label] (e.g., the particular animal) may be represented in each image of the second set of unlabeled data … and one or more of the parameters of each unlabeled example from the second set of unlabeled data 103 may be different than parameters in each labeled example from the first set of labeled data 101.” Note that the different parameters are the same as varying characteristics.);
obtaining augmented training data based on the supervised training data and the weakly labeled training data (Note that Taccari does not label the ‘augmented training data’, they simply say the ‘labeled and pseudo-labeled data sets,’ as seen below); and
training, via machine learning, a robust content categorization model based on the augmented training data. (Paragraph [0028], “Training (at 108) the model may include performing semi-supervised machine learning on the labeled and pseudo-labeled data sets 101 and 103,” see Fig. 1A Ref No. 108. Note that Taccari does not title the model as the ‘robust content categorization model’ but they are understood to be the same thing)
Regarding claim 2, Taccari teaches all of the limitations of the method in claim 1 as cited above and further teaches the generating process:
wherein the generating the weakly labeled training data comprises: training, via machine learning, a base pseudo label prediction model based on the supervised training data; (Paragraph [0046], “Process 300 may include inputting (at 306) a partitioned first group of unlabeled data to the mapping function and/or model trained with the set of labeled data.” Note that machine learning just implies any type of training and that Taccari refers to the ‘supervised training data’ as the ‘labeled data,’ as seen above. Further, note that Taccari does not title the model as the ‘base pseudo label prediction model’ but they are understood to be the same thing)
and creating the weakly labeled training data based on the unlabeled data clusters and the base pseudo label prediction model (Paragraph [0046], “The mapping function and/or model may classify each unlabeled data item from the first group of unlabeled data to one of the labels specified for the set of labeled data, ” Note that the ‘weakly labeled training data’ is understood to be the group of unlabeled data items classified with a label via the model, Taccari does not title this data group but they are understood to be the same thing, i.e. the output of the mapping function/model creates the ‘weakly labeled training data’ in Taccari)
Regarding claim 3, Taccari teaches all of the limitations of the method in claim 2 as cited above and further teaches the creating process:
Wherein the creating comprises: for each of the unlabeled data clusters, predicting, for each unlabeled data sample in the unlabeled data cluster, a pseudo label based on the base pseudo label prediction model, wherein the predicted pseudo label is one of the plurality of labels (Paragraph [0046], “The mapping function and/or model may classify each unlabeled data item from the first group of unlabeled data to one of the labels specified for the set of labeled data.” Again, note that the ‘base pseudo label prediction model’ from the present application is known as the model trained using the labeled data set, as is the case in the citation above)
determining a cluster label for the unlabeled data cluster based on the pseudo labels
predicted respectively for the unlabeled data samples in the unlabeled data cluster, wherein the cluster label if one of the plurality of labels (Paragraph [0047], “Process 300 may include defining (at 308) a particular pseudo-label for the partitioned first group of unlabeled data based on the label or classification that is output by the model for the greatest number of unlabeled data items from the particular first group of unlabeled data.” Note that although it doesn’t state it in the above citation, the pseudo-label for the partitioned first group must also be of the plurality of labels from the labeled data because the labels given to each individual data item, used to select the cluster label, had to have been one of the plurality of labels from the labeled data, as seen above), and
assigning the cluster label to each of the unlabeled data samples in the unlabeled
data cluster (Paragraph [0047], “In other words, defining (at 308) the particular pseudo-label may include overriding the classifications that are output by the model for some of the unlabeled data items.” Note that ‘overriding the classification’ is understood to mean the same thing as assigning the selected cluster label to each data item, regardless of its original label. Further, note that the classification is only overridden for some of the unlabeled data items because -some were already given the selected corresponding cluster label as their predicted label, therefore the cluster label is assigned to each of the unlabeled data samples in some manner); and
generating, based on the data samples in the unlabeled data clusters with newly assigned cluster labels, the weakly labeled training data (Paragraph [0047], “Process 300 may include defining (at 308) a particular pseudo-label for the partitioned first group of unlabeled data based on the label or classification that is output by the model,” Note that Taccari does not title the data group as the ‘weakly labeled training data’ but they are understood to be the same thing.)
Regarding claim 4, Taccari teaches all of the limitations of the method in claim 3 as cited above and further teaches the determining process:
wherein the determining a cluster label for the unlabeled data cluster comprises: processing the pseudo labels predicted for the unlabeled data samples in the unlabeled data cluster (Note that to use the pseudo-labels for anything, it can be understood that they have to be processed a.k.a. understood first); obtaining at least one metric associated with the pseudo labels for data samples in the unlabeled data cluster (Paragraph [0056], “Accordingly, model 403 may output (at 406) first set of probabilities 407 for each image of first subset of unlabeled data 405. Each probability from first set of probabilities 407 may specify a degree of certainty for classifying the input unlabeled data to one of several different classifications or labels,” Note that in the above citation the at least one metric is understood to be the probability or degree of certainty in classifying each particular input); and
selecting one of the pseudo labels as the cluster label according to a predetermined function based on the at least one metric (Paragraph [0057], “DSSL system 100 may aggregate first set of probabilities 407 that is output by model 403 for each image of first subset of unlabeled data 405. DSSL system 100 may evaluate the aggregated first set of probabilities 407 in order to determine the label that is used to classify the majority or greatest number of unlabeled data items from first subset of unlabeled data 405. DSSL system 100 may define (at 408) a particular pseudo-label for each unlabeled data item of the received set of unlabeled data based on the label used to classify the majority or greatest number of unlabeled data items from first subset of unlabeled data 405.” Note that the aggregation function is a predetermined function which makes use of the probability metric identified above)
Regarding claim 5, Taccari teaches all of the limitations of the method in claim 4 as cited above and further teaches the metric:
Wherein the at least one metric includes a confidence score for each of the pseudo labels, or a frequency for each of the pseudo labels (Paragraph [0047], “Process 300 may include defining (at 308) a particular pseudo-label for the partitioned first group of unlabeled data based on the label or classification that is output by the model for the greatest number of unlabeled data items from the particular first group of unlabeled data and/or based on whichever label or classification is identified with the greatest accuracy for the particular first group of unlabeled data.” Note that the output given for the greatest number of unlabeled data items is understood to be a frequency metric. Further, note that the greatest accuracy is understood to be a confidence score)
Regarding claim 6, Taccari teaches all of the limitations of the method of claim 4 as cited above and further teaches the use of the predetermined function:
Wherein the predetermined function (Paragraph [00002], the aggregate function) is used for identifying the cluster label based on one of the pseudo labels that has a maximum confidence score, or is associated with a highest frequency (Paragraph [0057], “DSSL system 100 may aggregate first set of probabilities 407 that is output by model 403 for each image of first subset of unlabeled data 405. DSSL system 100 may evaluate the aggregated first set of probabilities 407 in order to determine the label that is used to classify the majority or greatest number of unlabeled data items from first subset of unlabeled data.” Note that, as above, the greatest number of a certain pseudo-label is understood to be the same thing as a highest frequency of a certain pseudo-label. Further, note that you can see an example process using a maximum confidence score above and in paragraphs [0047]-[0049])
Regarding claim 7, Taccari teaches all of the limitations of the method in claim 1 as cited above and further teaches the process for using the trained model comprising:
receiving content to be categorized (Paragraph [0064], “Process 500 may include receiving (at 506) unlabeled data items from a particular sensor of a particular device. In particular, DSSL system 100 may receive a real-time feed of images generated by the particular sensor as the particular device is in motion” Note that given the present application, content is understood to be any type of text, audio, or visual information, of which Taccari teaches the latter in the above citation); and
classifying the content based on the robust content categorization model (Paragraph [0065], “Process 500 may include classifying (at 508) the unlabeled data items using the generated (at 504) model.” Note that, as above, the robust content categorization model was understood to be the model trained using the labeled and pseudo-labeled data sets; see the citation below for the description of the model at 504) trained based on the augmented training data including both the supervised training data and the weakly labeled training data (Paragraph [0063], “Process 500 may include generating (at 504) a model based on the DSSL performed over the received (at 502) labeled and unlabeled data. Once again, the unlabeled data items from a common domain may be grouped into an equivalence class and provided a common pseudo-label.” Note that as described throughout Taccari, the model based on the DSSL performed over the data is the one trained using labeled and pseudo-labeled data sets)
Regarding claim 8, see the rejection for claim 1 above. Note that the only difference between claim 8 and claim 1 is that claim 8 is directed to the article of manufacture whereas claim 1 is directed to the method that the article of manufacture carries out. Additionally, Taccari teaches that the operations and processes cited in the rejection for claim 1 may be in the form of a computer-readable medium. (Paragraph [0112], “Device 1000 may perform certain operations relating to one or more processes described above. Device 1000 may perform these operations in response to processor 1020 executing software instructions stored in a computer-readable medium”)
Regarding claim 9, see the rejection for claim 2 above. Note that the only difference between claim 9 and claim 2 is that claim 9 is directed to the article of manufacture whereas claim 2 is directed to the method that the article of manufacture carries out. See the rejection for claim 8 above.
Regarding claim 10, see the rejection for claim 3 above. Note that the only difference between claim 10 and claim 3 is that claim 10 is directed to the article of manufacture whereas claim 3 is directed to the method that the article of manufacture carries out. See the rejection for claim 8 above.
Regarding claim 11, see the rejection for claim 4 above. Note that the only difference between claim 11 and claim 4 is that claim 11 is directed to the article of manufacture whereas claim 4 is directed to the method that the article of manufacture carries out. See the rejection for claim 8 above.
Regarding claim 12, see the rejection for claim 5 above. Note that the only difference between claim 12 and claim 5 is that claim 12 is directed to the article of manufacture whereas claim 5 is directed to the method that the article of manufacture carries out. See the rejection for claim 8 above.
Regarding claim 13, see the rejection for claim 6 above. Note that the only difference between claim 13 and claim 6 is that claim 13 is directed to the article of manufacture whereas claim 6 is directed to the method that the article of manufacture carries out. See the rejection for claim 8 above.
Regarding claim 14, see the rejection for claim 7 above. Note that the only difference between claim 14 and claim 7 is that claim 14 is directed to the article of manufacture whereas claim 7 is directed to the method that the article of manufacture carries out. See the rejection for claim 8 above.
Regarding claim 15, see the rejection for claim 1 above. Note that the only difference between claim 15 and claim 1 is that claim 15 is directed to the system whereas claim 1 is directed to the method that the system carries out. Additionally, Taccari teaches that the operations and processes cited in the rejection for claim 1 may be performed by various computer components including, potentially, a training data augmenter and an augmented data-based model training engine. (Paragraph [0112], “Device 1000 may perform certain operations relating to one or more processes described above,” and Paragraph [0108], “FIG. 10 illustrates example components of device 1000. One or more of the devices described above may include one or more devices 1000. Device 1000 may include bus 1010, processor 1020, memory 1030, input component 1040, output component 1050, and communication interface 1060. In another implementation, device 1000 may include additional, fewer, different, or differently arranged components.”)
Regarding claim 16, see the rejection for claim 2 above. Note that the only difference between claim 16 and claim 2 is that claim 16 is directed to the system whereas claim 2 is directed to the method that the system carries out. See the rejection for claim 15 above.
Regarding claim 17, see the rejection for claim 3 above. Note that the only difference between claim 17 and claim 3 is that claim 17 is directed to the system whereas claim 3 is directed to the method that the system carries out. See the rejection for claim 15 above.
Regarding claim 18, see the rejection for claim 4 above. Note that the only difference between claim 18 and claim 4 is that claim 18 is directed to the system whereas claim 4 is directed to the method that the system carries out. See the rejection for claim 15 above.
Regarding claim 19, see the rejection for claims 5 and 6 above. Note that claim 19 recites the same limitations as claims 5 and 6 when combined with an “and.” Additionally, note that the only difference between claim 19 and claims 5 and 6 is that claim 19 is directed to the system whereas claims 5 and 6 are directed to the method that the system carries out. See the rejection for claim 15 above.
Regarding claim 20, see the rejection for claim 7 above. Note that the only difference between claim 20 and claim 7 is that claim 20 is directed to the system whereas claim 7 is directed to the method that the system carries out. Further, Taccari teaches the additional computer component introduced in claim 20, see the rejection for claim 15 above.
Citation of Pertinent Prior Art
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Altaf et al. (US 20220383127 A1) teaches a method for training a model to label a set of unlabeled nodes using a confidence score to select pseudo-labels for the set of nodes. The pseudo-labeled set of notes and labeled nodes are then combined to re-train the model. The end model is capable of classifying an unclassified node. Luong et al. (US 20220083840 A1) teaches a self-training method for creating a model that can be used in classification of data. The method first trains a model on labeled data. Then, the method creates a pseudo label for each item of an unlabeled data set using the trained model. These pseudo-labeled data and the labeled data are then combined to train another model. During training, noise is incorporated into the combined set of data to introduce variation in the features of the data.
Conclusion
The prior art made of record and not relied upon is considered pertinent to Applicant's disclosure. Applicant is required under 37 C.F.R. § 1.111(c) to consider these references fully when responding to this action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Zane A Rawlings whose telephone number is (571)270-3372. The examiner can normally be reached M-F, 8am to 5pm 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, Alexey Shmatov can be reached at (571) 270-3428. 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.
/Z.A.R./Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights