Notice of Pre-AIA or AIA Status
1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
2. This Office Action is sent in response to Applicant’s Communication received on 08/24/2026 for application number 18/384,525.
Response to Amendments
3. The Amendment filed 08/24/2026 has been entered. Claims 1-20 remain pending in the application.
Response to Arguments
The 101 rejection:
With respect to Step 2A, Prong two, the Examiner has considered claim 1 as a whole consistent with Ex Parte Desjardins. The Examiner acknowledges that the present Specification describes a technical resource problem associated with conventional federated HDC training and identifies embodiments that may reduce computational, communication, energy, and storage requirements. However, claim 1 does not recite the particular resource efficient training arrangement on which Applicant relies. Claim 1 merely requires processing circuitry to “train one or more independent sub models of the HDC model” and “transmit the trained one or more independent sub models … to another computing device.” Accordingly, unlike Desjardins and Enfish, claim 1 does not sufficiently reflect the technological mechanism alleged to provide the asserted improvement. The generic memory, processing circuitry, and transmission to another computing device therefore do not integrate the judicial exception into a practical application.
With respect to Step 2B, Applicant’s reliance on BASCOM is also unpersuasive. The additional elements have been considered both individually and as an ordered combination. Claim 1 merely requires generic memory and processing circuitry to train the HDC sub-models and transmit the trained sub-models to another computing devices. Thus, the ordered combination does not provide the particular non-conventional technological arrangement found sufficient in BASCOM. Rather, the additional elements amount to generic computer implementation and communication of the result of the recited model training activity.
Claim Rejections:
Applicant argues that Sharad's disclosure of a "locally trained model Li”. However, a "sub model" is a structural proper subset or portion of the overall model. Sharad explicitly discloses that each client derives its local model Li "by training the current global model G^(r-1) with their own training data" (Sharad, paragraph [0014]). Thus, Sharad's local model is a full-sized, structurally identical instance of the global model that is trained on a local subset of data; it is not a structural "sub model" of the global model itself. Examiner respectfully disagrees and notes that the claims do not recite that a “sub model” must be a structural proper subset of the dimensions, parameters, or other internal elements of the overall model. Under, BRI, “sub model” encompasses a separately trained model or model component associated with and used as part of the overall model training arrangement. Moreover, the rejection does not rely upon Sharad alone for the complete HDC structure. Xu teaches the HDC model and separately identifiable class/prototype related model components, while Sharad teaches independently training model information at respective client devices and transmitting the locally trained model information to another computing device for aggregation. The references are lied upon for their combined teachings.
Applicant argues that Salamat merely teaches matching the i-th dimension of a query vector with the i-th dimension of a class vector (i.e., a "corresponding dimension") to compute a cosine similarity dot product. Salamat does not teach that a dimension itself corresponds to an entire prototype (e.g., a class hypervector). Examiner respectfully disagrees and notes that the claim does not state that an individual dimension must be, equal, or be uniquely mapped one-to-one to an entire prototype. It merely recites that each dimension “corresponding to” a prototype. Salamat teaches representative class hypervectors that serve as class/prototype representations and teaches operations at individual dimensional level with respect to those prototype representations. In particular, paragraph [0036] teaches multiplying each dimension of the query vector by the corresponding dimension of the class hypervectors. Thus, Salamat establishes a correspondence between the dimensional components of the HDC representation and the prototype/class hypervector representations.
Applicant argues that Claim 9 requires initializing a prototype that is a sub-model and iteratively training the initialized prototype, whereas Sharad trains a complete local model. Examiner respectfully disagrees and notes that Sharad is not required to independently teach every aspect of claim 9. Xu is relied upon for the HDC specific training structure, including prototype/class hypervector formation and retraining, while Sharad is relied upon for transmission to another computing device. Accordingly, Xu teaches an HDC class/prototype representation that is initialized and subsequently retrained, such a prototype reasonably corresponds to the claimed sub-model. Sharad’s teaching of locally and independently training model information and transmitting the trained results to another computing device to be combined with Xu’s HDC prototype training.
Claim Rejections - 35 USC § 101
4. 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 the abstract idea without significantly more.
Step 1, the claims are directed to a process, machine, and manufacture.
Step 2A Prong 1, Claims 1, 9, and 16 recite, in part
the HDC model having a number of dimensions and a number of prototypes, each dimension of the HDC model corresponding to a protype of the number of prototypes (Mathematical concepts, mathematical relationships).
Step 2A Prong 2, this judicial exception is not integrated into a practical application.
The additional elements:
memory; and processing circuitry (mere instructions to apply the exception using a generic computer component).
transmit the trained one or more independent sub models of the HDC model to another computing device (mere data gathering and output recited at a high level of generality, and thus are insignificant extra-solution activity).
Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception, either alone or in combination.
The additional elements:
memory; and processing circuitry (mere instructions to apply the exception using a generic computer component).
transmit the trained one or more independent sub models of the HDC model to another computing device (mere data gathering and output recited at a high level of generality, and thus are insignificant extra-solution activity).
Claims 2-8, 10-15, 17-20 provide further limitations to the abstract idea (Mathematical concepts and/or Mental processes) as rejected in claims 1, 9, 16, however, they do not disclose any additional elements that would amount to a practical application or significantly more than an abstract idea (data gathering/insignificant extra-solution activity and/or generic computer component).
Claim Rejections – 35 USC § 103
5. 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 of this title, 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.
6. Claims 1, 5-9, 13-16, and 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Xu et al. (U.S. Patent Application Pub. No. US 20230409871 A1) in view of Sharad et al. (U.S. Patent Application Pub. No. US 20200285980 A1), and further in view of Salamat et al. (U.S. Patent Application Pub. No. US 20210334703 A1).
Claim 1: Xu teaches a device to train a hyperdimensional computing (HDC) model (i.e. hyperdimensional computing (HDC) for classification has been emerging as a lightweight machine learning framework targeting inference on resource-constrained edge devices. HDC classifiers mimic the brain cognition process by representing an object as a vector (e.g., a hypervector) with a very high dimension … the training process of an HDC classifier is extremely simple—simply averaging over the hypervectors of labeled training samples to derive the corresponding class hypervectors; para. [0042, 0045]), comprising:
memory (i.e. memory; para. [0049]); and
processing circuitry (i.e. processor; para. [0049]) to:
train one or more sub models of the HDC model (i.e. Given K classes, the training process obtains K class hypervectors 205, each for one class; para. [0060, 0061]), the HDC model having a number of dimensions (i.e. HDC classifiers mimic the brain cognition process by representing an object as a vector (e.g., a hypervector) with a very high dimension. The dimension can be on the order of thousands of bits, and potentially even higher numbers of bits; para. [0042, 0048, 0051]) and a number of prototypes (i.e. Given K classes, the training process obtains K class hypervectors 205, each for one class, re-training leads to an adjusted centroid for each class; para. [0060, 0061]), dimension of the HDC model corresponding to a protype of the number of prototypes (i.e. The HDC encodes all the values, features, and samples as hyperdimensional bipolar vectors, e.g., H∈{1, −1}D=10,000; para. [0051, 0056, 0061, 0074]); and
transmit the to another computing device (i.e. Client computer(s)/devices 50 and server computer(s) 60 provide processing, storage, and input/output devices executing application programs and the like. Client computer(s)/devices 50 can also be linked through communications network 70 to other computing devices, including other client devices/processes 50 and server computer(s) 60; para. [0104]).
Xu does not explicitly teach train one or more independent sub models; each dimension of the model corresponding to a protype; transmit the trained one or more independent sub models of the model to another computing device.
However, Sharad teaches train one or more independent sub models of the HDC model (i.e. the aggregated group updates are obtained from a locally trained model L_i using local training data by each participant; para. [0019, 0020]); and transmit the trained one or more independent sub models of the HDC model to another computing device (i.e. the selected clients send updates to the central server for improving the global model based on their local data; para. [0014, 0020, 0045]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Xu to include the feature of Sharad. One would have been motivated to make this modification because it enables distributed or federated training of the HDC model, to reduce centralized data collection and preserve local data privacy.
However, Salamat teaches each dimension of the HDC model corresponding to a protype of the number of prototypes (i.e. The similarity in the fixed-point and power-of-two number representations is defined as calculating the cosine similarity, which is obtained by multiplying each dimension in the query vector to the corresponding dimension of the class hypervectors, and adding up the partial products; para. [0005, 0034, 0036]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu and Sharad to include the feature of Salamat. One would have been motivated to make this modification because it provides a clearer dimension-wise correspondence between the HDC representation and the class representative vectors used for classification.
Claim 5: Xu, Sharad, and Salamat teach the device of claim 1. Xu further teaches wherein to train the one or more sub models of the HDC model, the processing circuitry is to: transform one or more training data points to one or more hyperdimensional representations (i.e. FIG. 2 is a block diagram 200 illustrating an example an overview of an existing HDC model. In this example, an input sample is represented as a vector 201 with N features F={f1, f2, . . . fN}, where the value range for each feature is normalized and uniformly discretized into M values, i.e., fi∈{1, . . . , M} for i=1, . . . , N. The HDC encodes all the values, features, and samples as hyperdimensional bipolar vectors; para. [0051, 59, 60]); initialize a prototype using the one or more hyperdimensional representations of the one or more training data points (i.e. Given K classes, the training process obtains K class hypervectors 205, each for one class; para. [0060]), wherein the initialized prototype is a sub-model of the one or more sub models of the HDC model (i.e. Given K classes, the training process obtains K class hypervectors 205, each for one class; para. [0060]); and iteratively train the initialized prototype (i.e. To improve accuracy, existing HDC models have also added re-training as part of the training process. Concretely, re-training fine tunes the class hypervectors C 205 derived ; para. [0061]).
Xu does not explicitly teach train the one or more independent sub models.
However, Sharad further teaches train the one or more independent sub models (i.e. the aggregated group updates are obtained from a locally trained model L_i using local training data by each participant; para. [0019, 0020]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu and Salamat to include the feature of Sharad. One would have been motivated to make this modification because it enables distributed or federated training of the HDC model, to reduce centralized data collection and preserve local data privacy.
Claim 6: Xu, Sharad, and Salamat teach the device of claim 5. Xu further teaches wherein to transform the one or more training data points to the one or more hyperdimensional representations comprises mapping the one or more training data points into a hyperdimensional vector (i.e. An encoding module 220 encodes input sample as a sample hypervector 204 by fetching the pre-generated value and feature hypervectors 202 and 203 from the item memory (IM) 206. Specifically, by combining each feature hypervector 203 with its corresponding value hypervector 202, the encoding output for an input sample is given by: S=sgn (Σi=1 NFiºVfi) where fi is the i-th feature value, Vfi is the corresponding value hypervector, is the Hadamard product, and sgn(·) is the sign function that binarizes the encoded sample hypervector. As a tiebreaker, sgn(0)=1; para. [0059, 0060]).
Xu does not explicitly teach using random projection-based HDC mapping.
However, Salamat further teaches using random projection-based HDC mapping (i.e. base hypervectors can be selected randomly; para. [0031-0033]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu and Sharad to include the feature of Salamat. One would have been motivated to make this modification because it provides a clearer dimension-wise correspondence between the HDC representation and the class representative vectors used for classification.
Claim 7: Xu, Sharad, and Salamat teach the device of claim 5. Xu further teaches wherein the prototype has one or more classes, the one or more training data points corresponding to a class of the prototype (i.e. Given K classes, the training process obtains K class hypervectors 205, each for one class; para. [0056, 0060, 0061]).
Claim 8: Xu, Sharad, and Salamat teach the device of claim 5. Xu further teaches wherein the processing circuitry is further to: for a transformed data point of the one or more transformed data points: determine a distance between the transformed data points to each prototype of the HDC model; compare distances between the transformed data points and the prototypes of the HDC model; and select a prototype with a smallest distance based on comparison of the distances (i.e. re-training leads to an adjusted centroid for each class. The most similar class hypervector 205 with the lowest Hamming distance indicates the classification result: arg min k Hamm(Sq, Ck). Due to the equivalence of Hamming distance and cosine similarity, the classification rule (arg mink Hamm(Sq, Ck)) is equivalent to arg maxk Sq T Ck which essentially transforms the bit-wise comparison to vector multiplication and is instrumental for establishing the equivalence between an HDC model and a neural network; para. [0061-0063]).
Claim 9: Xu teaches a method of training a hyperdimensional computing (HDC) model (i.e. Given K classes, the training process obtains K class hypervectors 205, each for one class … re-training fine tunes the class hypervectors C 205; para. [0060, 0061]) having a number of dimensions (i.e. HDC classifiers mimic the brain cognition process by representing an object as a vector (e.g., a hypervector) with a very high dimension. The dimension can be on the order of thousands of bits, and potentially even higher numbers of bits; para. [0042, 0048, 0051]) and a number of prototypes (i.e. Given K classes, the training process obtains K class hypervectors 205, each for one class, re-training leads to an adjusted centroid for each class; para. [0060, 0061]), dimension of the HDC model corresponding to a protype of the number of prototypes (i.e. The HDC encodes all the values, features, and samples as hyperdimensional bipolar vectors, e.g., H∈{1, −1}D=10,000; para. [0051, 0056, 0061, 0074]), the method comprising, on a first device (i.e. FIG. 1 is a block diagram 100 illustrating an example embodiment of an exemplary tiny device 101 for the inference process of the proposed LDC classifier, as disclosed herein. The processor 103 is coupled with an item memory (IM) 104 and an associative memory (AM) 105; para. [0049]):
transforming one or more training data points to one or more hyperdimensional representations (i.e. An encoding module 220 encodes input sample as a sample hypervector 204 … Training data 208 having data and annotations (e.g., a feature hypervector) is encoded into a sample vector S 205 using the encoding module 220; para. [0059, 0060]);
initializing a prototype using the one or more hyperdimensional representations of the one or more training data points (i.e. The training process averages the sample hypervectors 204 within a class using the equation Ck=sgn(ΣS∈Ω k S), where Ωk is the set of sample hypervectors 204 in k-th class. The resulting class hypervectors 205 generated by the training are stored in the associative memory (AM) 209; para. [0060]), wherein the initialized prototype is a sub-model of one or more independent sub models of the HDC model (i.e. Given K classes, the training process obtains K class hypervectors 205, each for one class; para. [0060]);
iteratively training the initialized prototype (i.e. To improve accuracy, existing HDC models have also added re-training as part of the training process. Concretely, re-training fine tunes the class hypervectors C 205 derived as C k=sgn(ΣS∈Ω k S). If a training sample 208 is misclassified, it is given more weights in correct class hypervector and subtracted from the wrong class hypervector. Essentially, re-training leads to an adjusted centroid for each class; para. [0061]); and
transmitting the to another computing device (i.e. Client computer(s)/devices 50 and server computer(s) 60 provide processing, storage, and input/output devices executing application programs and the like. Client computer(s)/devices 50 can also be linked through communications network 70 to other computing devices, including other client devices/processes 50 and server computer(s) 60; para. [0104]).
Xu does not explicitly teach each dimension of the model corresponding to a protype; transmit the trained one or more independent sub models of the model to another computing device.
However, Sharad teaches transmiting the trained one or more independent sub models to another computing device (i.e. the selected clients send updates to the central server for improving the global model based on their local data; para. [0014, 0020, 0045]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Xu to include the feature of Sharad. One would have been motivated to make this modification because it enables distributed or federated training of the HDC model, to reduce centralized data collection and preserve local data privacy.
However, Salamat teaches each dimension of the HDC model corresponding to a protype of the number of prototypes (i.e. The similarity in the fixed-point and power-of-two number representations is defined as calculating the cosine similarity, which is obtained by multiplying each dimension in the query vector to the corresponding dimension of the class hypervectors, and adding up the partial products; para. [0005, 0034, 0036]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu and Sharad to include the feature of Salamat. One would have been motivated to make this modification because it provides a clearer dimension-wise correspondence between the HDC representation and the class representative vectors used for classification.
Claims 13-15 are similar in scope to Claims 5-8 and are rejected under a similar rationale.
Claim 16 is similar in scope to Claim 1 and is rejected under a similar rationale.
Xu teaches at least one non-transitory computer-readable medium with instructions stored thereon (i.e. Memory 90 provides volatile storage for computer software instructions 92 and data 94 used to implement an embodiment of the present invention (e.g., the inference classification method, system, techniques, and program code detailed above). Disk storage 95 provides non-volatile storage for computer software instructions 92 and data 94 used to implement an embodiment of the present invention. Central processor unit 84 is also attached to system bus 79 and provides for the execution of computer instructions; para. [0105]), which, when executed by a processor (i.e. processor; para. [0049]) of a computing device, cause the processor to (i.e. FIG. 1 is a block diagram 100 illustrating an example embodiment of an exemplary tiny device 101 for the inference process of the proposed LDC classifier, as disclosed herein. The processor 103 is coupled with an item memory (IM) 104 and an associative memory (AM) 105; para. [0049]):
18-20 are similar in scope to Claims 5-8 and are rejected under a similar rationale.
7. Claims 2, 3, 10, 11, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Xu in view of Sharad, Salamat, and further in view of Chahal et al. (U.S. Patent Application Pub. No. US 20230409967 A1).
Claim 2: Xu, Sharad, and Salamat teach the device of claim 1. Xu does not explicitly teach wherein the another computing device is to: aggregate the trained one or more independent sub models with one or more trained additional sub models received from one or more additional devices; concatenate the aggregated sub models to create the model; and transmit the model to the device and the one or more additional devices.
However, Sharad further teaches wherein the another computing device is to: aggregate the trained one or more independent sub models with one or more trained additional sub models received from one or more additional devices (i.e. the aggregated group updates are obtained from a locally trained model L_i using local training data by each participant; para. [0019, 0020]); the aggregated sub models to create the HDC model (i.e. The server obtains aggregated group updates AU1, . . . , AUg from each group and compares the aggregated group updates and identifies suspicious aggregated group updates. The server combines the aggregated group updates by excluding the updates identified as suspicious, to obtain an aggregated update Ufinal. The server derives a new global model Gr from the previous model Gr-1 and the aggregated update Ufinal and shares Gr with the plurality of participants; para. [0005, 0018]); and transmit the HDC model to the device and the one or more additional devices (i.e. the server combines the aggregated group updates by excluding, or reducing the impact of, the updates identified as suspicious, obtaining an aggregated update Ufinal. At step 520, the server derives a new global model Gr from the previous model Gr-1 and the aggregated update Ufinal, and shares Gr with all participants; para. [0050]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu and Salamat to include the feature of Sharad. One would have been motivated to make this modification because it enables distributed or federated training of the HDC model, to reduce centralized data collection and preserve local data privacy.
However, Chahal teaches concatenate the aggregated sub models to create the model (i.e. concurrently reading and concatenating, the plurality of aggregated segment level DL models by each of the plurality of workers to obtain an aggregated model generated by the current training instance, wherein the aggregated model is used by each of the plurality of workers during the successive training instances using a successive mini batch; para. [0009]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu, Sharad, and Salamat to include the feature of Chahal. One would have been motivated to make this modification because it improves scalability in distributed training.
Claim 3: Xu, Sharad, Salamat, and Chahal teach the device of claim 2. Xu further teaches wherein the device and the one or more additional devices use the HDC model to perform an inference on test data (i.e. device 101 for the inference process of the proposed LDC classifier; para. [0049, 0062]) stored at the one or more additional devices to retrain at least a portion of the HDC model (i.e. retraining improves the inference accuracy against the basic HDC; para. [0091]).
Xu does not explicitly teach test data stored locally at the one or more additional devices.
However, Sharad further teaches wherein the device and the one or more additional devices use the HDC model to perform an inference on test data stored locally at the one or more additional devices (i.e. the system performs local training (LT). Each of the selected contributors Pi retrieves the current global model G, and proceeds with training a local model Li starting from G and using their own dataset D; para. [0019, 0039]) to retrain at least a portion of the HDC model (i.e. a federated learning protocol generates a global model (G) throughout several rounds of training; para. [0014]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu, Salamat, and Chahal to include the feature of Sharad. One would have been motivated to make this modification because it enables distributed or federated training of the HDC model, to reduce centralized data collection and preserve local data privacy.
Claims 10, 11, and 17 are similar in scope to Claims 2-3 and are rejected under a similar rationale.
8. Claims 4 and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Xu in view of Sharad, Salamat, Chahal, and further in view of Satheesh Kumar et al. (U.S. Patent Application Pub. No. US 20230297844 A1).
Claim 4: Xu, Sharad, Salamat, and Chahal teach the device of claim 2. Xu further teaches wherein at least one of the device or the one or more additional devices selects a subset of the HDC model to retrain the HDC model (i.e. retraining improves the inference accuracy against the basic HDC; para. [0091]) and transmits to the another computing device (i.e. Client computer(s)/devices 50 and server computer(s) 60 provide processing, storage, and input/output devices executing application programs and the like. Client computer(s)/devices 50 can also be linked through communications network 70 to other computing devices, including other client devices/processes 50 and server computer(s) 60; para. [0104]).
Xu does not explicitly teach selects a random subset of the model, transmits an updated model to the another computing device.
However, Sharad further teaches wherein at least one of the device or the one or more additional devices selects a subset of the HDC model to retrain the HDC model (i.e. a federated learning protocol generates a global model (G) throughout several rounds of training; para. [0014]) and transmits an updated HDC model to the another computing device (i.e. the selected clients send updates to the central server for improving the global model based on their local data; para. [0014, 0020, 0045]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu, Salamat, and Chahal to include the feature of Sharad. One would have been motivated to make this modification because it enables distributed or federated training of the HDC model, to reduce centralized data collection and preserve local data privacy.
However, Satheesh Kumar teaches wherein at least one of the device or the one or more additional devices selects a random subset of the model to retrain the model and transmits an updated model to the another computing device (i.e. the central computing device or server 102 may transmit to the users a global model (e.g., newly initialized or partially trained through previous rounds of federated learning). The users 104 may train their individual models locally with their own data. The results of such local training may then be reported back to central computing device or server 102, which may pool the results and update the global model. This process may be repeated iteratively. Further, at each round of training the global model, central computing device or server 102 may select a subset of all registered users 104 (e.g., a random subset) to participate in the training round; para. [0036]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Xu, Salamat, Sharad and Chahal to include the feature of Satheesh Kumar. One would have been motivated to make this modification because it reduces training burden while still allowing iterative refinement of the overall model.
Claim 12 is similar in scope to Claims 4 and is rejected under a similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure.
Zhang et al. (Pub. No. US 20250265475 A1), this application discloses a model training method, applied to the federated learning field, including: obtaining a plurality of first gradients and a plurality of second gradients; selecting some gradients from the plurality of first gradients and the plurality of second gradients, where the some gradients are used to update the target model in a current round of iteration of federated learning.
THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee 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 date of this final action.
It is noted that any citation to specific pages, columns, lines, or figures in the prior art references and any interpretation of the references should not be considered to be limiting in any way. A reference is relevant for all it contains and may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art. In re Heck, 699 F.2d 1331, 1332-33, 216 U.S.P.Q. 1038, 1039 (Fed. Cir. 1983) (quoting In re Lemelson, 397 F.2d 1006, 1009, 158 U.S.P.Q. 275, 277 (C.C.P.A. 1968)).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to TAN TRAN whose telephone number is (303)297-4266. The examiner can normally be reached on Monday - Thursday - 8:00 am - 5:00 pm MT.
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, Matt Ell can be reached on 571-270-3264. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/TAN H TRAN/Primary Examiner, Art Unit 2141
