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 .
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1 - 20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to abstract ideas without significantly more.
Regarding Claim 1:
Step 2A Prong 1:
based on predictions generated by the ML model in response to input data received by the ML model, labeling selected data of the input data; (Mental process, simply labelling input data can be done in the mind by observation.)
computing a set of metrics based on performance of the updated ML model; (Mathematical concepts, computing metrics is calculations merely based on math and can be done using pen and paper.)
Compiling the set of metrics into a reliable model exchange and aggregation score (RMEAS); and (Mental process, reviewing metrics and assigning a generic score can be done by the human mind)
Based on the RMEAS, determining whether to engage in an exchange, with one or more of the clients, involving the updated ML model (Mental process, simply choosing to exchange data or not can be done in the mind by judging the RMEAS.)
Step 2A Prong 2:
A method, compromising:
at a client that is a part of a group of clients, where each client in the group is able to communicate with the other clients of the group, performing operations compromising:
training a machine learning (ML) model hosted at the client (Adding insignificant extra-solution activity to the judicial exception MPEP 2106.05(g) – Examiner’s note: merely training a ML model with a dataset is generic and insignificant extra-solution activity.)
retraining the ML model, to create an updated ML model, using the selected data; (Adding insignificant extra-solution activity to the judicial exception MPEP 2106.05(g) – Examiner’s note: merely training a ML model with a dataset is generic and insignificant extra-solution activity.)
Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exceptions.
at a client that is a part of a group of clients, where each client in the group is able to communicate with the other clients of the group, performing operations compromising:
training a machine learning (ML) model hosted at the client (Training a generic ML model is well understood routine and conventional activity that adds nothing meaningful to the exceptions, see MPEP 2106.05(d))
retraining the ML model, to create an updated ML model, using the selected data; (Training a generic ML model is well understood routine and conventional activity that adds nothing meaningful to the exceptions, see MPEP 2106.05(d))
The additional elements as disclosed above alone or in combination do not integrate the judicial exceptions into a practical application as they are mere insignificant extra solution activity and applications of exceptions in combination.
This claim is ineligible.
Regarding claim 2:
The method as recited in claim 1, wherein the selected data is labelled by an oracle
This provides a further description of the method discussed with regard to claim 1. The limitations of claim 2, under broadest reasonable interpretation, does not include new abstract ideas.
Claim 2 recites additional elements of adding insignificant extra-solution activity to the judicial exception MPEP 2106.05(g) – (Examiner’s note: merely mentions who labelled the data. The labeler has no effect on the invention.)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exceptions when taken alone or in combination with previous claims. The claimed additional element is just insignificant extra-solution activity. Mentioning who labelled the data, when taken individually or in combination with previous additional elements, is well understood and routine conventional activity that adds nothing meaningful to the exceptions. Therefore, the judicial exceptions
are not integrated into a practical application.
Claim 2 is ineligible
Regarding Claim 3:
The method as recited in claim 1, wherein computing a set of metrics comprises:
determining an uncertainty measure corresponding to the predictions;
determining whether drift has occurred in the input data provided to the ML model; and
determining a reliability measure indicating whether labels of the selected data match with the predictions.
This provides a further description of the method discussed with regard to claim 1. The limitations of claim 3, under broadest reasonable interpretation, does include new abstract ideas of mental processes. Determining an uncertainty measure can be done in the mind by mentally evaluating the ML models’ predictions. Determining if drift occurred can be done in the mind by mentally judging the uncertainty measure. Determining a reliability measure can be done in the mind by counting when the data matches the predictions and by evaluating the matches into a score.
The claim does not include any additional elements that integrate the abstract idea into a practical application or amount to significantly more than the judicial exceptions, see MPEP 2106.05.
Claim 3 is ineligible.
Regarding claim 4:
The method as recited in claim 1, wherein determining whether to engage in the exchange involving the updated model compromises sharing the RMEAS with one or more of the other clients and/or comparing the EMEAS with a respective RMEAS received from one or more of the other clients.
This provides a further description of the method discussed with regard to claim 1. The limitations of claim 4, under broadest reasonable interpretation, does not include new abstract ideas.
Claim 4 recites additional elements of adding insignificant extra-solution activity to the judicial exception MPEP 2106.05(g) – (Examiner’s note: merely sharing data between generic computers.)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exceptions when taken alone or in combination with previous claims. The claimed
additional element is just insignificant extra-solution activity. Sharing data, when taken individually or in combination with previous additional elements, is well understood and routine conventional activity that adds nothing meaningful to the exceptions. Therefore, the judicial exceptions are not integrated into a practical application.
Claim 4 is ineligible
Regarding claim 5:
The method as recited in claim 1, wherein determining whether to engage in the exchange involving the updated ML model is based on an agreement protocol between the client and one or more of the other clients.
This provides a further description of the method discussed with regard to claim 1. The limitations of claim 5, under broadest reasonable interpretation, does include a new abstract idea of a method for organizing human activity. An agreement protocol is just an abstract interaction between the clients.
The claim does not include any additional elements that integrate the abstract idea into a practical application or amount to significantly more than the judicial exceptions.
Claim 5 is ineligible
Regarding claim 6:
The method as recited in claim 1, wherein when a determination is made to engage in the exchange, the client receives an external ML model from one of the other clients.
This provides a further description of the method discussed with regard to claim 1. The limitations of claim 6, under broadest reasonable interpretation, does not include new abstract ideas.
Claim 2 recites additional elements of adding insignificant extra-solution activity to the judicial exception MPEP 2106.05(g) – (Examiner’s note: merely receiving model data and parameters.)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exceptions when taken alone or in combination with previous claims. The claimed
additional element is just insignificant extra-solution activity. Receiving model data and parameters, when taken individually or in combination with previous additional elements, is well understood and routine conventional activity that adds nothing meaningful to the exceptions. Therefore, the judicial exceptions are not integrated into a practical application.
Claim 6 is ineligible
Regarding claim 7:
The method as recited in claim 6, wherein the client performs an evaluating process that comprises using the data to evaluate a performance of the external ML model.
This provides a further description of the method discussed with regard to claim 6. The limitations of claim 7, under broadest reasonable interpretation, does include a new abstract idea of a mental process. Evaluating a model’s performance can be done in the mind by observation and evaluation.
The claim does not include any additional elements that integrate the abstract idea into a practical application or amount to significantly more than the judicial exceptions.
Claim 7 is ineligible.
Regarding claim 8:
The method as recited in claim 7, wherein when the evaluating process indicates that the performance of the external ML model meets one or more established criteria, the external model is aggregated together with the update model to produce an aggregated model that is then deployed at the client.
This provides a further description of the method discussed with regard to claim 7. The limitations of claim 8, under broadest reasonable interpretation, does include a new abstract idea of a mathematical process. Aggregating 2 models together can be done mathematically.
Claim 8 recites additional elements of adding insignificant extra-solution activity to the judicial exception MPEP 2106.05(g) – (Examiner’s note: merely using a ML model on a generic computer.)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exceptions when taken alone or in combination with previous claims. The claimed
additional element is just insignificant extra-solution activity. Deploying a model, when taken individually or in combination with previous additional elements, is well understood and routine conventional activity that adds nothing meaningful to the exceptions. Therefore, the judicial exceptions are not integrated into a practical application.
Claim 8 is ineligible
Regarding claim 9:
The method as recited in claim 1, wherein an artifact is generated that comprises the input data to the ML model and the predictions generated based on the input data, and the artifact is provided to a drift detection module which determines whether drift has occurred in the ML predictions and/or in the input data.
This provides a further description of the method discussed with regard to claim 1. The limitations of claim 10, under broadest reasonable interpretation, does include a new abstract idea of a mental process. Determining if drift occurred can be done in the mind by observation and judgment.
Claim 9 recites additional elements of adding insignificant extra-solution activity to the judicial exception MPEP 2106.05(g) – (Examiner’s note: artifact made by gathering input data and prediction data. Also sending the artifact to the module is just generic transmitting of data).
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exceptions when taken alone or in combination with previous claims. The claimed
additional elements are just insignificant extra-solution activity. Gathering data and sending artifact data, when taken individually or in combination with previous additional elements, are well understood and routine conventional activity that adds nothing meaningful to the exceptions. Therefore, the judicial exceptions are not integrated into a practical application.
Claim 9 is ineligible
Regarding claim 10:
The method as recited in claim 1, wherein no local data of the nodes is exchanged during performance of the operations.
This provides a further description of the method discussed with regard to claim 1. The limitations of claim 10, under broadest reasonable interpretation, does not include new abstract ideas.
Claim 10 recites additional elements of adding insignificant extra-solution activity to the judicial exception MPEP 2106.05(g) – (Examiner’s note: merely specifies data used for the operations.)
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exceptions when taken alone or in combination with previous claims. The claimed
additional element is just insignificant extra-solution activity. Specifying that the data used for the operations can’t be local data, is well understood and routine conventional activity that adds nothing meaningful to the exceptions. Therefore, the judicial exceptions are not integrated into a practical application.
Claim 10 is ineligible
Regarding claims 11 – 20:
Claims 11 – 20 refer to a non-transitory storage medium which does the same activities as listed in the limitations of claims 1 – 10. Therefore, they receive the same judicial exceptions and additional elements as claims 1 – 10 where the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exceptions.
Claims 11 – 20 are ineligible.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1 – 8, 10 – 18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Ashour et al. (US 20240064793 A1) in view of Tamir et al. (US 20220036201 A1).
Regarding Claim 1,
Ashour teaches:
A method, compromising:
at a client that is a part of a group of clients, where each client in the group is able to communicate with the other clients of the group, performing operations compromising:
[68] teaches User Equipment (UE) making clusters or groups of users and forming nodes as clients. Fig 12 and [151] teaches a peer-to-peer federated learning network with no leader UE or central node.
training a machine learning (ML) model hosted at the client;
[201] teaches participant nodes of clients training local ML models on data. [208] also teaches each client or node trains and initializes their own ML model.
retraining the ML model, to create an updated ML model, using the selected data;
[202] teaches the participant nodes training their ML models again using selected data and updated weights. [143] also teaches retraining an ML model based on the same or different local data which is selected.
computing a set of metrics based on performance of the updated ML model;
[216] teaches performance-based metrics obtained from a peer node for an updated ML model. [214] and [215] teaches a performance metric can consist of a loss function to be minimized, a classification accuracy, a logarithmic loss, confusion matrix, or other evaluation metric.
compiling the set of metrics into a reliable model exchange and aggregation score (RMEAS); and
[211] and [216] teach interpreting one of the above-mentioned metrics as a score to signal model updates are reliable, can be exchanged, and aggregated. An RMEAS under its broadest reasonable interpretation is a score signaling a model is reliable, can be exchanged, and can be aggregated together with other models. The compiled score functions the same as an RMEAS and is compared to a threshold to make a decision.
based on the RMEAS, determining whether to engage in an exchange, with one or more of the clients, involving the update ML model
[216] teaches the performance metrics of the updated ML model being a requirement to share or refrain from sharing the ML model weights with other peer nodes.
Ashour doesn’t teach based on predictions generated by the ML model in response to input data received by the ML model labeling selected data of the input data;
Tamir teaches based on predictions generated by the ML model in response to input data received by the ML model labeling selected data of the input data” ([0063] teaches labelling labeled instances, which comes from an input data stream [0015]. [0005] teaches an example of input data and a missing data label to be labelled later. [0086] also teaches unlabeled data from the input. [0094] teaches reliable labels are discovered from the data and used to label other unlabeled data. Fig 5 and [0022] teaches a knowledge discovery module that obtains reliable data labels from a window of input data based on applications of machine learning models. The data from the window being a combination of labelled and unlabeled instances. The reliable labels obtained are the predicted labels for a data window given a labeling count and cost.)
It would have been obvious for one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine Ashour’s teaching with Tamir’s teaching. of labeling input data with reference to ML model predictions. This would allow further training of the machine learning model using estimated labelled data from previously unknown data to improve its performance metrics.
Regarding claim 2:
Ashour doesn’t teach The method as recited in claim 1, wherein the selected data is labelled by an oracle.
Tamir teaches The method as recited in claim 1, wherein the selected data is labelled by an oracle. ([0060] teaches manually labelled data by an expert which is the same as an oracle labelling data.)
Regarding Claim 3,
Ashour doesn’t teach:
The method as recited in claim 1, wherein computing a set of metrics comprises:
determining an uncertainty measure corresponding to the predictions;
determining whether drift has occurred in the input data provided to the ML model; and
determining a reliability measure indicating whether labels of the selected data match with the predictions.
Tamir teaches:
“determining an uncertainty measure corresponding to the predictions;” ([0087] teaches posterior probabilities generated from previous reliable labelled data predictions. This is an applicable uncertainty measure.)
“determining whether drift has occurred in the input data provided to the ML model; and” ( [102], [103], and Fig 7 teach how a decision is made whether drift has occurred in the input data or not by checking if a drift threshold has been crossed.)
“determining a reliability measure indicating whether labels of the selected data match with the predictions.” (Figure 8B teaches an accuracy measure for correctly labeled data based on their predicted labels. Fig 10B also teaches accuracy measures based on real and synthetic datasets in [0146]. The accuracy measure is determined to be a reliability measure.)
Regarding claim 4,
Ashour teaches:
The method as recited in claim 1, wherein determining whether to engage in the exchange involving the updated model compromises sharing the RMEAS with one or more of the other clients and/or comparing the RMEAS with a respective RMEAS received from one or more of the other clients.
Using the same rationale as claim 1, [216] teaches sharing a performance metric as an RMEAS. The metric being above a pre-configured threshold is the determining factor for whether to share updated ML model information with peer nodes or other clients. [217] also teaches the requesting node aggregating model updates together which can include comparing the performance metrics of multiple peer nodes.
Regarding claim 5,
Ashour teaches:
The method as recited in claim 1, wherein determining whether to engage in the exchange involving the updated ML model is based on an agreement protocol between the client and one or more of the other clients.
For the same reasons as claim 4, [216] teaches a pre-configured threshold for the model’s performance metric as the agreement protocol for determining eligibility to share model information. A peer node’s performance metric being above a threshold would permit sharing of model updates. The opposite would make the peer node refrain from sharing model updates.
Regarding claim 6,
Ashour teaches:
The method as recited in claim 1, wherein when a determination is made to engage in the exchange, the client receives an external ML model from one of the other clients.
[216] teaches the client receiving updated model information of an external model‘s modified weights residing with a different client or peer node.
Regarding claim 7,
Ashour teaches:
The method as recited in claim 6, wherein the client performs an evaluating process that comprises using the data to evaluate a performance of the external ML model.
For the same reasons as claim 5, [216] teaches an evaluation process of the performance metric being above a pre-configured threshold.
Regarding claim 8,
Ashour teaches:
The method as recited in claim 7, wherein when the evaluating process indicates that the performance of the external ML model meets one or more established criteria, the external model is aggregated together with the update model to produce an aggregated model that is then deployed at the client.
For the same reasons as claim 7, [216] teaches the evaluating process’s established criteria being a pre-configured threshold to pass. Once confirmed, [217] teaches the external models updated model information is aggregated together to produce a new updated model or aggregated model. The training of the aggregated model can be terminated afterwards and left for deployment at the individual node. [233] and [242] teaches deploying of the models by outputting sidelink communication parameters as the output of the aggregated ML models at each client node or UE.
Regarding claim 10,
Ashour teaches:
The method as recited in claim 1, wherein no local data of the nodes is exchanged during performance of the operations.
[209] teaches providing modified weights, percentages, or factors of weights that get shared. [139] and [145] teach data privacy and security by not sharing raw data from each node’s individual datasets.
Regarding Claims 11 – 18, and 20:
Claims 11 – 18 and 20 are machine claims with the same limitations as claims 1 – 8 and 10, and are rejected for the same reasons.
Claims 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Ashour et al. (US 20240064793 A1) in view of Tamir et al. (US 20220036201 A1) in view of Wenchel et al. (US 11922280 B2)
Regarding claim 9:
The method as recited in claim 1, … the input data to the ML model and the predictions generated based on the input data and … is provided to a drift detection module which determines whether drift has occurred in the ML predictions and/or in the input data.
Ashour over Tamir teaches this. Tamir [0094] teaches the drift detection method uses the received training input data. Tamir [0112] also teaches using reliable labelled data and posterior probabilities associated with the predictions in the
Ashour over Tamir does not teach “wherein an artifact is generated that comprises” and “the artifact”
Wenchel does teach “wherein an artifact is generated that comprises” and “the artifact” (Wenchel [Col 6, Line 46 - 60] teaches joint metrics or artifacts which are input and metric bundles that are generated from the input data and metrics based on them.)
It would have been obvious for one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine Ashour over Tamir’s teaching of input data and resulting ML predictions with Wenchel’s teaching of an artifact consisting of that data. The combination of the data has no effect on the resulting output of the drift module. Combining the data into one structure may allow faster and more efficient data processing since the input format is set and consistent across various trials. Processing one structure instead of a bundle of data may make data processing faster but the resultant determinations are the same.
Claim 19 is a machine claim with the same limitations as claim 9 and are rejected for the same reasons.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to AJAY J KANJOOR whose telephone number is (571)270-0965. The examiner can normally be reached Monday-Friday 8am-4pm.
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, Mariela Reyes can be reached at (571) 270-1006. 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.
/Mariela Reyes/Supervisory Patent Examiner, Art Unit 2142
Prosecution Insights