DETAILED ACTION
This Office Action is in response to the amendments filed on 06/24/2026.
Claims 1-5, 7, 8, 14-17, and 19 are currently amended.
Claims 6 and 20 are currently canceled.
Claims 21 and 22 are newly added.
Claims 1-5, 7-19, 21, and 22 are currently pending in this application and have been examined.
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 .
Allowable Subject Matter
Claims 5, 7, 8, 15, 16, objected to as being dependent upon a rejected base claim, but would be allowable over the prior art if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Claim Objections
Claim 1 objected to because of the following informalities: Claim 1 recites in part "generating validated local data . Appropriate correction is required.
Response to Arguments
In response to Applicant’s arguments on page(s) 13-17 regarding rejections made under 35 U.S.C. 101:
Claims 1-20 are rejected under 35 U.S.C. § 101 as allegedly being directed to non- statutory subject matter.
For at least the reasons discussed during the interview and without acquiescing in the rejection, the amended independent claims, and the claims that depend thereon, are patent- eligible under 35 U.S.C. § 101.
Similar to Example 39, the pending independent claims only recite features that are not practically be performed in the human mind. These features involve training a local model for federated learning, which is analogous to the claim of Examine 39 involving training a neural network.
Since the independent claims do not recite any abstract ideas for similar reasons as Example 39, the claims are patent-eligible under Step 2A, Prong One.
The claimed invention improves federated machine learning by: (1) preventing training a local model using inaccurate training data by corroborating client device data through comparison with external validation data; (2) improving the accuracy of the global model by excluding local models trained on unvalidated data from updates; and (3) preserving privacy by using de-identification functions to generate non-sensitive metadata by one or more of using a range to randomize sensitive information in the event metadata or removing multiple portions of the sensitive information in the event metadata while still enabling validation. See, for example, paragraphs [0002]-[0004] of the specification.
The amended independent claims reflect the disclosed improvement. For example, amended independent claim 1 recites "generating, using a data de-identification function and based on the event metadata, non-sensitive metadata by one or more of using a range to randomize sensitive information in the event metadata or removing multiple portions of the sensitive information in the event metadata" and "updating the global model based on the local model based on a comparison of the non-sensitive metadata and the corroborating metadata indicating that the local model should not be excluded from updates to the global model." These features of amended claim 1 provide an improvement to the technical field of federated machine learning by ensuring the accuracy and integrity of training data used to update global models while maintaining privacy through specific de-identification techniques.
Examiner’s response:
Applicant’s arguments have been fully considered but are found to be not persuasive.
Applicant argues that the instant application is similar to Example 39 of the Subject Matter Eligibility Examples. Examiner disagrees. Example 39 was found to be patent eligible because it does not recite any abstract ideas, even though some of the claim limitations recite actions that are based on mathematical concepts, namely the limitations regarding applying transformations to digital images. The instant application is distinct from that of Example 39 in that the instant application recites limitations that can be reasonably performed in the human mind, including generating event metadata, generating validated local data via exclusion of non-validated data, and generating deidentified data via anonymizing sensitive information in the data. The actions of anonymizing data and generating a dataset based on the anonymized data can be reasonably performed in the human mind.
Applicant argues that the instant application provides an improvement over the state of the art of federated learning. Examiner agrees partly. While the amended limitations of independent Claim 1 recite an improvement over the art of federated learning in that the global model is updated based on a comparison of anonymized data and the corroborating metadata, the parallel independent claims 3 and 17 do not recite such an improvement, and as such the rejections made under 35 U.S.C. 101 as applied to independent Claim 1 are withdrawn. It is the recommendation of the Examiner that the parallel independent claims be amended to recite similar limitations.
In light of the amendments made to the claims, the rejections made under 35 U.S.C. 101 are maintained and updated below.
In reference to Applicant’s arguments on page(s) 17 regarding rejections made under 35 U.S.C. 103:
Claims 1-4, 6, 10-14, and 17, and 19 are rejected under 35 U.S.C. § 103 as allegedly being unpatentable over JOSHI (US Patent Application Publication No. 2022/0201021) in view of COSMAN (US Application Publication No. 2021/0192078), and claims 9 and 18 are rejected under 35 U.S.C. § 103 as allegedly being unpatentable over JOSHI and COSMAN as applied to claims 1, 3 and 17 above, and further in view of SONG (US Patent No. 8,856,051).
For at least the reasons agreed to during the interview and without acquiescing in the rejection, the cited sections of the applied references, whether taken alone or in any reasonable combination, do not disclose one or more features recited in the amended independent claims. Therefore, the amended independent claims and the claims that depend thereon are patentable over the cited sections of the applied references, whether taken alone or in any reasonable combination.
Accordingly, Applicant respectfully requests that the Examiner reconsider and withdraw this rejection.
Examiner’s response:
Applicant’s arguments have been fully considered and are found to be persuasive.
Applicant argues that the applied references of Joshi and Cosman do not teach the amended claims. Examiner agrees in part. Joshi and Cosman are both silent on the use of external validation metadata to obtain information pertaining to device information of a user and information in the anonymized metadata that corresponds to the user device information. This deficiency is not remedied by the inclusion of Song. However, these limitations are only present in independent Claim 17, and as such the art rejections for claim 17 are withdrawn and the remaining claims are still rejected.
In light of the amendments made on the claims, the rejections made under 35 U.S.C. 103 are maintained and updated below.
Claim Rejections - 35 USC § 101
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 2-5, 7-19, 21, and 22 rejected under 35 U.S.C. 101 because they are directed to an abstract idea without significantly more.
Step 1 analysis:
Independent Claim 1 recites, in part, a system for federated learning, therefore falling into the statutory category of manufacture. Independent Claim 3 recites, in part, a method for federated learning, therefore falling into the statutory category of process. Independent Claim 17 recites, in part, a non-transitory, computer-readable medium comprising instructions recorded thereon, therefore falling into the statutory category of manufacture.
Regarding Claim 2:
Step 2A: Prong 1 analysis:
Claim 2 recites in part:
“identifying a first plurality of attributes in the non-sensitive metadata”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying attributes in a dataset.
“identifying a second plurality of attributes in the corroborating metadata”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying attributes in a dataset.
“determining not to exclude the local model from updates to the global model based on whether a number of first attributes from the first plurality of attributes that match second attributes from the second plurality of attributes is below a threshold”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses making a determination on excluding model updates based on a threshold value.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 3:
Step 2A: Prong 1 analysis:
Claim 3 recites in part:
“generating, based on first validation data received from a first external server, event metadata corresponding to events in the local data, wherein the first validation data is related to the events in the local data, and wherein the first validation data is inaccessible to a central server”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses generating metadata about events, metadata comprising data like location of event, time of event, type of event, length of event, etc.
“based on comparing the local data and the event metadata, determining to exclude a portion of the local data that is not validated by the event metadata, and generating validated local data”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses determining to exclude certain data.
“generating, using a data de-identification function, and based on the event metadata, de-identified event metadata by one or more of using a range to randomize sensitive information in the event metadata or removing multiple portions of the sensitive information in the event metadata”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses generating non-sensitive data via randomization or deletion.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The judicial exception is not integrated into practical application. In particular, the claim recites the additional elements of:
“accessing local data to be used for training a local model on a client device”. This additional element is recited at a high level of generality and amounts to extra-solution activity of gathering data i.e. pre-solution activity of gathering data for use in the claimed process.
“training the local model based on the validated local data”. This limitation merely indicates a field of use or technological environment in which the judicial exception is performed (federated/distributed learning) and thus fails to add an inventive concept to the claims. See MPEP 2106.05(h).
“transmitting the local model and the de-identified event metadata to the central server”. This additional element is recited at a high level of generality and amounts to extra-solution activity of gathering data i.e. pre-solution activity of gathering data for use in the claimed process.
Accordingly at Step 2A: Prong 2, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
The additional element(s) of “accessing local data to be used for training a local model on the client device” and “transmitting the local model and the de-identified event metadata to the central server” is/are recited at a high level of generality and amount(s) to extra-solution activity of receiving data i.e., pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
The additional element(s) of “training the local model based on the validated local data” is/are directed to particular field(s) of use (federated/distributed learning) (MPEP 2106.05(h)) and therefore do not provide significantly more than the abstract idea, and thus the claim is subject-matter ineligible.
Accordingly, at Step 2B, the additional elements individually or in combination do not amount to significantly more than the judicial exception.
Regarding Claim 4:
Step 2A: Prong 1 analysis:
Claim 4 recites in part:
“identifying sensitive information in the event metadata, wherein the sensitive information comprises information that is capable of being used to identify a user associated with the client device”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying sensitive data.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 5:
Step 2A: Prong 1 analysis:
Claim 5 recites in part:
“determining that the sensitive information should be randomized”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses determining to randomize identifying data.
“generating non-sensitive metadata by randomizing the sensitive information in the event metadata, wherein randomizing the sensitive information in the event metadata comprises generating one or more values within a predetermined range to replace the sensitive information in the event metadata”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses anonymizing data by randomization.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 7:
Step 2A: Prong 1 analysis:
Claim 7 recites in part:
“determining that the sensitive information should be removed”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses determining to anonymize identifying data.
“generating non-sensitive metadata by removing the multiple portions of the sensitive information in the event metadata”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses anonymizing data by way of removal.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 8:
Step 2A: Prong 1 analysis:
Claim 8 recites in part:
“generating a sensitivity threshold, wherein the sensitivity threshold is based on a combination of information in the local data that is capable of being used to identify a user associated with the client device”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses creating a threshold based on analyzing data.
“determining a sensitivity metric, wherein the sensitivity metric is based on an amount of information contained in a subset of the local data that is capable of being used to identify the user associated with the client device”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses creating a metric based on analyzing data.
“determining that the sensitivity metric meets or exceeds the sensitivity threshold”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying that a threshold value is met.
“excluding the subset of the local data with the sensitivity metric exceeding the sensitivity threshold”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses withholding data that does not meet a threshold value.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 9:
Step 2A: Prong 1 analysis:
Claim 9 recites in part:
“identifying a first attribute in the event metadata”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying data.
“identifying a second attribute in the local data”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying data.
“comparing the first attribute and the second attribute by searching for a similarity between the first attribute and the second attribute that indicates the second attribute is accurate”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses finding a similarity between data points.
“determining that the first attribute does not validate the second attribute by determining that the first attribute and the second attribute do not share the similarity”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses determining that two data points are not similar.
“excluding unvalidated data from the local data used to train the local model”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses withholding data that is not validated.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 10:
Step 2A: Prong 2 analysis:
The judicial exception is not integrated into practical application. In particular, the claim recites the additional elements of:
“wherein transmitting the local model and the de-identified event metadata to the central server comprises transmitting the local model and the de-identified event metadata to an aggregator, wherein the aggregator validates and sends the local model and the de-identified event metadata to the central server to integrate with a global model, and wherein the validation comprises generating the validated local data”. This additional element is recited at a high level of generality and amounts to extra-solution activity of gathering data i.e. pre-solution activity of gathering data for use in the claimed process.
Accordingly at Step 2A: Prong 2, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
The additional element(s) of “wherein transmitting the local model and the de-identified event metadata to the central server comprises transmitting the local model and the de-identified event metadata to an aggregator, wherein the aggregator validates and sends the local model and the de-identified event metadata to the central server to integrate with a global model, and wherein the validation comprises generating the validated local data” is/are recited at a high level of generality and amount(s) to extra-solution activity of receiving data i.e., pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Accordingly, at Step 2B, the additional elements individually or in combination do not amount to significantly more than the judicial exception.
Regarding Claim 11:
Step 2A: Prong 1 analysis:
Claim 11 recites in part:
“wherein generating an average comprises averaging the first parameters of the local model and the second parameters of the second local model”. As drafted and under its broadest reasonable interpretation, this limitation covers a mathematical calculation.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The judicial exception is not integrated into practical application. In particular, the claim recites the additional elements of:
“wherein the aggregator generates an average of first parameters of the local model with second parameters of a second local model from a second client device and transmits the average to the central server for integration with the global model”. This additional element is recited at a high level of generality and amounts to extra-solution activity of gathering data i.e. pre-solution activity of gathering data for use in the claimed process.
Accordingly at Step 2A: Prong 2, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
The additional element(s) of “wherein the aggregator generates an average of first parameters of the local model with second parameters of a second local model from a second client device and transmits the average to the central server for integration with the global model” is/are recited at a high level of generality and amount(s) to extra-solution activity of receiving data i.e., pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Accordingly, at Step 2B, the additional elements individually or in combination do not amount to significantly more than the judicial exception.
Regarding Claim 12:
Step 2A: Prong 1 analysis:
Claim 12 recites in part:
“combines first weights of the local model and second weights of the second local model using a weighted average, wherein using the weighted average comprises taking the weighted average of the first weights of the local model and the second weights of the second local model”. As drafted and under its broadest reasonable interpretation, this limitation covers a mathematical calculation.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The judicial exception is not integrated into practical application. In particular, the claim recites the additional elements of:
“wherein the aggregator collects the local model from the client device, collects a second local model from a second client device”. This additional element is recited at a high level of generality and amounts to extra-solution activity of gathering data i.e. pre-solution activity of gathering data for use in the claimed process.
“and transmits the weighted average to the central server for integration with the global model”. This additional element is recited at a high level of generality and amounts to extra-solution activity of gathering data i.e. pre-solution activity of gathering data for use in the claimed process.
Accordingly at Step 2A: Prong 2, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
The additional element(s) of “wherein the aggregator generates an average of first parameters of the local model with second parameters of a second local model from a second client device and transmits the average to the central server for integration with the global model” and “and transmits the weighted average to the central server for integration with the global model” is/are recited at a high level of generality and amount(s) to extra-solution activity of receiving data i.e., pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Accordingly, at Step 2B, the additional elements individually or in combination do not amount to significantly more than the judicial exception.
Regarding Claim 13:
Step 2A: Prong 2 analysis:
The judicial exception is not integrated into practical application. In particular, the claim recites the additional elements of:
“wherein the aggregator resides on the central server and the central server performs aggregation of the local model with local models from other client devices and updates the global model”. This additional element is recited at a high level of generality and amounts to extra-solution activity of gathering data i.e. pre-solution activity of gathering data for use in the claimed process.
Accordingly at Step 2A: Prong 2, the additional elements individually or in combination do not integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
The additional element(s) of “wherein the aggregator resides on the central server and the central server performs aggregation of the local model with local models from other client devices and updates the global model” is/are recited at a high level of generality and amount(s) to extra-solution activity of receiving data i.e., pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Accordingly, at Step 2B, the additional elements individually or in combination do not amount to significantly more than the judicial exception.
Regarding Claim 14:
Step 2A: Prong 1 analysis:
Claim 14 recites in part:
“identifying relevant information from the first validation data, wherein the relevant information comprises application data residing outside the client device that comprises data related to portions of data on the client device”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying relevant information in a dataset.
“generating the event metadata pertaining to the application data on the client device corresponding to the relevant information”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses generating metadata about data, metadata comprising data like location of event, time of event, type of event, length of event, etc.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 15:
Step 2A: Prong 1 analysis:
Claim 15 recites in part:
“identifying relevant information from the first validation data, wherein the relevant information comprises metadata pertaining to device information”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying relevant information in a dataset.
“generating the event metadata pertaining to the relevant information”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses generating metadata about events, metadata comprising data like location of event, time of event, type of event, length of event, etc.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 16:
Step 2A: Prong 1 analysis:
Claim 16 recites in part:
“identifying relevant information from the first validation data, wherein the relevant information comprises metadata pertaining to communication data”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying relevant information in a dataset.
“generating the event metadata pertaining to the relevant information”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses generating metadata about events, metadata comprising data like location of event, time of event, type of event, length of event, etc.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 17:
Due to claim language similar to that of Claims 1 and 3, Claim 17 is rejected for the same reasons as presented above in the rejection of Claims 1 and 3, with the exception of the limitations covered below.
Step 2A: Prong 1 analysis:
Claim 17 recites in part:
“identifying relevant information from the validation data, wherein the relevant information comprises metadata pertaining to device information”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying relevant information in a dataset.
“identifying data in the de-identified event metadata corresponding to the relevant information”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses identifying data from anonymized data.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 18:
Due to claim language similar to that of Claim 9, Claim 18 is rejected for the same reasons as presented above in the rejection of Claim 9.
Regarding Claim 19:
Due to claim language similar to that of Claim 14, Claim 19 is rejected for the same reasons as presented above in the rejection of Claim 14.
Regarding Claim 21:
Step 2A: Prong 1 analysis:
Claim 21 recites in part:
“generating the de-identified event metadata by using the range to randomize the sensitive information in the event metadata”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses randomizing sensitive information.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Regarding Claim 22:
Step 2A: Prong 1 analysis:
Claim 22 recites in part:
“generating the de-identified event metadata by removing the multiple portions of the sensitive information in the event metadata”. As drafted and under its broadest reasonable interpretation, this limitation covers performance of the limitation in the mind (including an observation, evaluation, judgement, or opinion) or with the aid of pencil and paper. For example, this limitation encompasses removing sensitive information.
Accordingly, at Step 2A: Prong 1, the claim is directed to an abstract idea.
Step 2A: Prong 2 analysis:
The claim does not recite any additional elements that integrate the judicial exception into a practical application.
Step 2B analysis:
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception.
Claim Rejections - 35 USC § 103
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claim(s) 1-4, 10-14, 17, 19, and 21 is/are rejected under 35 U.S.C. 103 as being unpatentable over Joshi et al (US 20220201021 A1, hereinafter Joshi) in view of Cosman et al (US 20210192078 A1, hereinafter Cosman).
Regarding Claim 1:
Joshi teaches
and wherein the first validation data is inaccessible to a central server (Joshi [0016]: "In particular, and as described in further detail below, examples disclosed herein prevent global models from being updated with local model data in circumstances where client-side model data validation (e.g., stage 1 validation) has not occurred. For example, in the event a particular client is compromised, model parameters corresponding to that client may be poisoned in an effort to weaken and/or otherwise disrupt the efficacy of a global model derived from any number of local client models.")
generating validated local data by exclude a portion of the local data that is not validated by the event metadata (Joshi [0016]: "In particular, and as described in further detail below, examples disclosed herein prevent global models from being updated with local model data in circumstances where client-side model data validation (e.g., stage 1 validation) has not occurred. For example, in the event a particular client is compromised, model parameters corresponding to that client may be poisoned in an effort to weaken and/or otherwise disrupt the efficacy of a global model derived from any number of local client models.");
training the local model based on the validated local data (Joshi [0016]: "Example federated learning approaches disclosed herein enable ML models existing in a client's local cloud to be validated (e.g., stage 1 validation) after training and before sending tokenized data to a server");
generating, using a data de-identification function, and based on the event metadata, non-sensitive metadata by one or more of using a range to randomize sensitive information in the event metadata or removing multiple portions of the sensitive information in the event metadata (Joshi [0018]: "In some examples, a system tokenizes large samples of network telemetry data to avoid communicating any explicit telemetry information, and such tokenized data only includes information relevant for training a ML model. In this example, the data after tokenization contains no PII or data that would allow PII to be derived, while still containing information that is relevant to training a ML model.");
the central server comprising one or more second processors and a second non-transitory, computer-readable medium storing second instructions that, when executed by the one or more second processors, cause second operations comprising (Joshi [0081]: "The processor platform 1500 of the illustrated example also includes one or more mass storage devices 1528 for storing software and/or data"; (EN): it can be seen in Fig. 15 that the server architecture includes a processor 1512 coupled to memory and storage devices)
generating, based on second validation data received from a second external server, corroborating metadata corresponding to the events in the non- sensitive metadata, wherein the second validation data is related to the events in the non-sensitive metadata, and wherein the second validation data is inaccessible to the client device; (Joshi [0023]: "the server-side validation manager 106 generates global ML models updated by processing the data sent by any number of the client-side validation managers 110. In some examples the outputs of a global ML model existing in the server cloud 102 are validated by the server-side validation manager 106 and, if valid (e.g., if metrics corresponding to model accuracy satisfy one or more thresholds), sends model updates to the client-side validation manager 110 to update a local model with a global model. Validity is found by inputting ground truth data (e.g., known to be legitimate) into the model and comparing the model's outputs to ground truth outputs.");
updating the global model based on the local model based on a comparison of the non-sensitive metadata and the corroborating metadata indicating that the local model should not be excluded from updates to the global model (Joshi [0086]: “perform a secondary validation corresponding to the client-side model parameters using a server-side ground truth data set, and determine whether to update the global model with the client-side model parameters based on a comparison of results of the secondary validation and a validation threshold”).
Joshi does not distinctly disclose
A system for federated learning including validation of training data on client devices before training models based on the training data, the system comprising: a client device, one or more first processors, and a first non-transitory, computer-readable medium storing first instructions that, when executed by the one or more first processors, cause first operations comprising:
accessing local data to be used for training a local model on the client device;
generating event metadata corresponding to events in the local data, wherein the first validation data is related to the events in the local data
based on first validation data received from a first external server
transmitting the local model and the non-sensitive metadata to the central server;
receiving, from the client device, the local model and the non-sensitive metadata;
However, Cosman teaches
A system for federated learning including validation of training data on client devices before training models based on the training data, the system comprising: a client device, one or more first processors, and a first non-transitory, computer-readable medium storing first instructions that, when executed by the one or more first processors, cause first operations comprising: (Cosman [0090]: "FIG. 8 is a block diagram of a device architecture 800 for a mobile or embedded device, according to an embodiment. The device architecture 800 includes a memory interface 802, a processing system 804 including one or more data processors, image processors and/or graphics processing units")
accessing local data to be used for training a local model on the client device (Cosman [0031]: "the model parameters θ are transmitted to each device and then each device trains a local model using its local data. The locally trained model on device i becomes θ.sub.i and the difference between the starting model parameters and the locally trained parameters Δ.sub.i=θ.sub.i−θ are sent to a central server.");
generating event metadata corresponding to events in the local data, wherein the first validation data is related to the events in the local data (Cosman [0030]: "The features learned may be features and/or metadata associated with the content item and/or usage patterns related to the content item"; [0052]: "Knowledge graph local data 384 may indicate metadata associated with the content item, including, but not limited to the following: location, person presence (e.g., contact presence, coworker presence, family member presence, partner presence, child presence, etc.), animal presence (e.g., pet, wildlife, etc.), category for content item (e.g., public event, business, scene type, home, work, etc.), period of time (e.g., taken on weekend, taken during summer, taken during a particular season or holiday, taken during week, etc.), an indication on return on investment, and any other knowledge graph information on the content item.")
based on first validation data received from a first external server (Cosman [0055]: "The method 400 includes operations (block 401) to transmit a machine learning model from a server to a set of client devices. Let θ∈Θ be model parameters for a particular model that is to be trained using user data. The model parameters θ are transmitted to each device in a set of client devices.")
transmitting the local model and the non-sensitive metadata to the central server (Cosman [0046]: "The local machine learning modules 136a-136n, 137a-137n, 138a-138n on each client device can generate model updates that are privatized by the client devices 110a-110n, 111a-111n, 112a-112n before transmission to the server 130. For example, client devices 110a-110n can each send privatized model updates 121, client devices 111a-111n can each send privatized model updates 122, and client devices 112a-112n can each send privatized model updates 123. The privatized model updates can be sent through the network 120 to the server 130, where the updates can be processed into an aggregated model update 135. Updates are sent to the server while satisfying separated differential privacy for the local updates and no raw data for users is transmitted to the server.");
receiving, from the client device, the local model and the non-sensitive metadata (Cosman [0040]: "FIG. 1 illustrates a system 100 to enable a user behavioral model developed with private federated learning, according to an embodiment. In one embodiment the system 100 includes a server 130 that can receive data from a set of client devices 110a-110n, 111a-111n, 112a-112n over a network 120"; (EN): it can be seen in Cosman that the server receives data from the local models, and the transmission of data can be combined with the anonymized data of Joshi in order to transmit both the local models and the anonymized data);
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to combine the methods for privacy preserving models of Joshi with the method for private federated learning of Cosman in order to provide a method for local models of a federated learning system to obscure any personal data of users via differential privacy. The method presented in Cosman is beneficial for Joshi in that it allows for private federated learning to or update a machine learning model to predict user behavior with items accessible with a device, while maintaining user privacy as data gets transmitted to a central server (Cosman [0033]: “One approach for incorporating privacy into federated learning is to use differential privacy on the server, typically called the central model of differential privacy. Differential privacy has many nice properties, including closure under post-processing and the graceful degradation of privacy parameters if multiple differential privacy algorithms are composed together, that has made it the de facto privacy definition in data analytics and machine learning”).
Regarding Claim 2:
Joshi teaches
The system of claim 1, wherein the second operations further comprise: identifying a first plurality of attributes in the non-sensitive metadata (Joshi [0016]: "Unlike raw telemetry data that may contain information deemed sensitive and/or otherwise personal, tokenized data removes indications of personally identifiable information (PII). In some examples, tokenized data includes model parameters that include no indication of labelling.");
identifying a second plurality of attributes in the corroborating metadata (Joshi [0023]: "In some examples the server-side validation manager 106 generates global ML models updated by processing the data sent by any number of the client-side validation managers 110. In some examples the outputs of a global ML model existing in the server cloud 102 are validated by the server-side validation manager 106 and, if valid (e.g., if metrics corresponding to model accuracy satisfy one or more thresholds), sends model updates to the client-side validation manager 110 to update a local model with a global model");
and determining not to exclude the local model from updates to the global model based on whether a number of first attributes from the first plurality of attributes that match second attributes from the second plurality of attributes is below a threshold (Joshi [0025]: "The classification is added to the tokenized message as a flag (e.g., a signal, bit, data, etc.), that signifies whether the sent data is valid or invalid, after the client-side validation occurs in the client-side validation manager 110. The classification being valid if ground truth data stored in the local data storage 108, after being run through the trained model in the example client cloud 104, passes with a confidence metric over a configurable threshold. The classification being invalid if the ground truth data stored in the local data storage 108, after being run through the trained model in the example client cloud 104, does not pass with a confidence metric over the configured threshold. Using this information, the example data aggregator 204 discards data flagged as invalid and averages the data that was flagged as valid").
Regarding Claim 3:
Due to claim language similar to that of Claim 1, Claim 3 is rejected for the same reasons as presented above in the rejection of Claim 1.
Regarding Claim 4:
Joshi teaches
The method of claim 3, wherein generating the de-identified event metadata comprises: identifying sensitive information in the event metadata, wherein the sensitive information comprises information that is capable of being used to identify a user associated with the client device (Joshi [0016]: "Unlike raw telemetry data that may contain information deemed sensitive and/or otherwise personal, tokenized data removes indications of personally identifiable information (PII)"; [0021]: “Instead, clients and ISPs alike desire competent services and/or protection without relinquishing so much data that might be deemed private (e.g., telemetry data having labels that indicates port numbers, device manufacturers, device models, time-of-use, etc.)”);
Regarding Claim 10:
Joshi teaches
The method of claim 3, wherein transmitting the local model and the de-identified event metadata to the central server comprises transmitting the local model and the de-identified event metadata to an aggregator, wherein the aggregator validates and sends the local model and the de-identified event metadata to the central server to integrate with a global model, and wherein the validation comprises generating the validated local data (Joshi [0025]: "The example data aggregator 204 receives tokenized data from any number of clients corresponding to a same model being trained in any number of client clouds 104. In some examples all tokenized data sent to the data aggregator 204 is flagged with a validation classification... Using this information, the example data aggregator 204 discards data flagged as invalid and averages the data that was flagged as valid. The example server training manager 206 receives and/or otherwise retrieves the output of the averaged data to train a global model. In some examples, both invalid and valid data received by the data aggregator 204 is utilized.").
Regarding Claim 11:
Joshi teaches
The method of claim 10, wherein the aggregator generates an average of first parameters of the local model with second parameters of a second local model from a second client device and transmits the average to the central server for integration with the global model, wherein generating an average comprises averaging the first parameters of the local model and the second parameters of the second local model (Joshi [0025]: "Using this information, the example data aggregator 204 discards data flagged as invalid and averages the data that was flagged as valid. The example server training manager 206 receives and/or otherwise retrieves the output of the averaged data to train a global model. In some examples, both invalid and valid data received by the data aggregator 204 is utilized."; [0064]: "The example server-side model accuracy calculator 212 calculates accuracy metrics of the client-side model parameter data (block 906), and if one or more accuracy thresholds is satisfied, the local model parameters are averaged (block 908). The example server-side model validator 210 updates and/or otherwise authorizes the updating of the global model in connection with the averaged parameter data in an effort to improve the overall efficacy of the global model").
Regarding Claim 12:
Joshi teaches
The method of claim 10, wherein the aggregator collects the local model from the client device, collects a second local model from a second client device, combines first weights of the local model and second weights of the second local model using a weighted average, wherein using the weighted average comprises taking the weighted average of the first weights of the local model and the second weights of the second local model, and transmits the weighted average to the central server for integration with the global model (Joshi [0030]: "In some examples, if the local model is validated (e.g., the accuracy value is above the threshold) by the example client-side model validator 312, the example client reporting manager 306 communicates the new model by sending tokenized data. The tokenized data may include, but is not limited to, any number of weights of the ML model (e.g., input weights, hidden layer weights, and/or output weights), tokenized telemetry data, training information (e.g., time it took to train, accuracy value, etc.), and/or model type so the data aggregator 204 can gather all information from any number of client-side validation managers 110 pertaining to the same model type."; [0064]: "The example server-side model accuracy calculator 212 calculates accuracy metrics of the client-side model parameter data (block 906), and if one or more accuracy thresholds is satisfied, the local model parameters are averaged (block 908). The example server-side model validator 210 updates and/or otherwise authorizes the updating of the global model in connection with the averaged parameter data in an effort to improve the overall efficacy of the global model").
Regarding Claim 13:
Joshi teaches
The method of claim 10, wherein the aggregator resides on the central server and the central server performs aggregation of the local model with local models from other client devices and updates the global model (Joshi [0025]: "The example data aggregator 204 receives tokenized data from any number of clients corresponding to a same model being trained in any number of client clouds 104. In some examples all tokenized data sent to the data aggregator 204 is flagged with a validation classification... Using this information, the example data aggregator 204 discards data flagged as invalid and averages the data that was flagged as valid. The example server training manager 206 receives and/or otherwise retrieves the output of the averaged data to train a global model. In some examples, both invalid and valid data received by the data aggregator 204 is utilized.").
Regarding Claim 14:
Joshi teaches
The method of claim 3, wherein generating event metadata corresponding to the events in the local data comprises: identifying relevant information from the first validation data, wherein the relevant information comprises application data residing outside the client device that comprises data related to portions of data on the client device (Joshi [0029]: "In some examples, the local data storage 108 categorizes and stores telemetry data produced by IoT devices 112 connected to the client network. Telemetry data is categorized in relation to its type (e.g., device ID, UPnP, mDNS, DNS, IP flow data, and TLS metadata, ground truth, etc.) and therefore can be accessed by category relevant to a specific model and/or task.");
and generating the event metadata pertaining to the application data on the client device corresponding to the relevant information (Joshi [0023]: "the server-side validation manager 106 generates global ML models updated by processing the data sent by any number of the client-side validation managers 110. In some examples the outputs of a global ML model existing in the server cloud 102 are validated by the server-side validation manager 106 and, if valid (e.g., if metrics corresponding to model accuracy satisfy one or more thresholds), sends model updates to the client-side validation manager 110 to update a local model with a global model. Validity is found by inputting ground truth data (e.g., known to be legitimate) into the model and comparing the model's outputs to ground truth outputs.").
Regarding Claim 21:
Joshi teaches
The method of claim 3, wherein generating the de-identified event metadata comprises: generating the de-identified event metadata by removing the multiple portions of the sensitive information in the event metadata (Joshi [0018]: "In some examples, a system tokenizes large samples of network telemetry data to avoid communicating any explicit telemetry information, and such tokenized data only includes information relevant for training a ML model. In this example, the data after tokenization contains no PII or data that would allow PII to be derived, while still containing information that is relevant to training a ML model.").
Claim Rejections - 35 USC § 103
Claim(s) 9 is/are rejected under 35 U.S.C. 103 as being unpatentable over Joshi and Cosman as applied to claims 1 and 3 above, and further in view of Song et al (US 8856051 B1, hereinafter Song).
Regarding Claim 9:
Joshi teaches
The method of claim 3, wherein determining to exclude the portion of the local data that is not validated by the event metadata further comprises: identifying a first attribute in the event metadata (Joshi [0016]: "Unlike raw telemetry data that may contain information deemed sensitive and/or otherwise personal, tokenized data removes indications of personally identifiable information (PII). In some examples, tokenized data includes model parameters that include no indication of labelling.");
identifying a second attribute in the local data (Joshi [0030]: "For example, if a local model was trained for device identification, the example local data storage 108 might contain a list of known devices with identification information about each device");
and excluding unvalidated data from the local data used to train the local model (Joshi [0016]: "In particular, and as described in further detail below, examples disclosed herein prevent global models from being updated with local model data in circumstances where client-side model data validation (e.g., stage 1 validation) has not occurred.").
Joshi does not distinctly disclose
comparing the first attribute and the second attribute by searching for a similarity between the first attribute and the second attribute that indicates the second attribute is accurate;
determining that the first attribute does not validate the second attribute by determining that the first attribute and the second attribute do not share the similarity;
However, Song teaches
comparing the first attribute and the second attribute by searching for a similarity between the first attribute and the second attribute that indicates the second attribute is accurate (Song [Col 5 lines 41-44]: "The metadata augmentation system 119 comprises a similarity determination module 210 that determines similarities between digital objects, such as the videos 117 of FIG. 1.");
determining that the first attribute does not validate the second attribute by determining that the first attribute and the second attribute do not share the similarity (Song [Col 9 line 65-Col 10 line 5]: "Similarly, classifier output indicating that a video in the negative validation set is in fact similar is considered a false positive (FP), output indicating that a video in the positive validation set is not similar is considered a false negative (FN), and output indicating that a video in the negative validation set is not similar is considered a true negative (TN). An overall accuracy can then be determined for the classifier, with classifiers lacking sufficient accuracy being discarded."; (EN): a true negative output is similar to the first attribute not validating the second attribute);
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to combine the methods for privacy preserving models of Joshi + Cosman with the methods for a metadata augmentation system that determines similarities between digital objects of Song in order to provide a method for validation of metadata attributes via comparing the attributes to each other. The method presented in Song is beneficial for Joshi + Cosman in that it allows for a federated learning model to easily search for similarities or differences between the local models and the global model in order to more accurately validate and train the model(s) (Song [Col 1 lines 15-19]: “Accurate labeling of a video is of great value in such systems, permitting users to search for videos corresponding to given labels, the video hosting service itself to train classifiers for video concepts using the label metadata, and the like”).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 20220360450 A1 – one or more of anonymizing, via an anonymization service hosted within a trusted execution environment (TEE), raw data provided by a computing node to generate anonymized data
US 20230102732 A1 – Systems and methods facilitate privacy-preserving data curation in a federated learning system
US 20220351860 A1 – A method for using a federated learning classifier in digital pathology
US 20220044162 A1 – blockchain-based secure federated learning
US 20210406782 A1 – A system for decentralized federated learning
US 11188791 B2 – A computer-implemented method for training a global federated learning model using an aggregator server
US 20200082290 A1 – Techniques that facilitate adaptive anonymization of data using statistical inference
THIS ACTION IS MADE FINAL. 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 COREY M SACKALOSKY whose telephone number is (703)756-1590. The examiner can normally be reached M-F 7:30am-3:30pm EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Omar Fernandez Rivas can be reached at (571) 272-2589. 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.
/COREY SACKALOSKY/Examiner, Art Unit 2128
/OMAR F FERNANDEZ RIVAS/Supervisory Patent Examiner, Art Unit 2128
Prosecution Insights