DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Remarks
This communication is considered fully responsive to the Amendment filed on 6/26/26.
Response to Arguments
Applicant's arguments filed 6/26/26 have been fully considered but they are not persuasive.
1] applicant argues (summary Remarks 6/26/26, pg 8-9) (emphasis added)
Independent claim 1, as amended, sets forth instructions including "analyze one or more properties of the first endpoint, the second endpoint, and the third endpoint to select the first locally trained model and the second locally trained model for aggregation" and "aggregate the first locally trained model and the second locally trained model to produce a new central model, the third model excluded from the new central model." The Beser/Tan/Blanchard combination fails to teach or suggest such instructions.
...
Because each of Beser, Tan, and Blanchard is missing the same elements of claim 1, namely, analyzing one or more properties of the endpoints to select models for aggregation, with the third model excluded from the new central model, the alleged Beser/Tan/Blanchard combination is missing those same elements. Therefore, the Beser/Tan/Blanchard combination fails to establish a prima facie case of obviousness of the instructions of claim 1.
Withdrawal of the §103 rejections of claim 1 and all claims dependent thereon is respectfully
requested.
The examiner respectfully disagrees.
Specifically, Beser, Tan and Blanchard disclose aggregate the first locally trained model and the second locally trained model to produce a new central model, (Blanchard: fig 1-10, [0015-132]: fig 1 … a first computer “parameter server” and n worker computers similar to a general distributed system model … a portion f of the workers are possible “Byzantine” i.e. they may deliver erroneous and/or arbitrary results [0049] … deep learning solution deployed and trained over several computers (1ST 2ND 3RD ... n endpoints) and parameter vector is a vector comprising all the synaptic weights and the internal parameters of the deep learning model (1ST 2ND 3RD ... n locally trained models) … cost function is any measure of deviation between what deep learning model predicts and what the computers actually observe [0058] … the present method operates only on a subset (i.e. exclude certain models) of the received estimate vectors (aggregate the first locally trained model and the second locally trained model to produce a new central model) [0017]),
the third model excluded from the new central model (Blanchard: fig 1-10, [0015-132]: fig 1 … determining the updated parameter vector precludes (excludes) the estimate vectors which have a distance greater than a predefined maximum distance to the other estimate vectors (see with [0049;58;17] above - the third model excluded from the new central model) … the first computer does not aggregate all estimate vectors received from worker computers but disregards (excludes) vectors that are too far away from the other vectors, sinch such outliers are erroneous with high likelihood (see with [0049;58;17] above - the third model excluded from the new central model) [0018]).
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 1-20 are rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1-10 and 12-18 of U.S. Patent No. 12169584. Although the claims at issue are not identical, they are not patentably distinct from each other because conflicting claims are in a patent by the same inventive entity. Furthermore, where claims in the instant application are broader than the claims of the ‘ 584 patent, it would have been obvious to one of ordinary skill in the art at the time the invention was made to omit elements when the remaining elements perform as before. A person of ordinary skill could have arrived at the present claims by omitting the details of the ‘584 patent claims. See In re Karlson (CCPA) 136 USPQ 184, decided January 16, 1963 ("Omission of element and its function in combination is obvious expedient if remaining elements perform same function as before").
Instant application
‘584 patent
Claims 1, 8, 15 (claim 1 exemplary)
At least one non-transitory computer-readable storage medium comprising instructions to cause at least one processor circuit to at least:
distribute a central model to a first endpoint, a second endpoint, and a third endpoint;
access a first locally trained model, the first locally trained model created by training the central model at the first endpoint using first local data, the first local data local to the first endpoint;
access a second locally trained model, the second locally trained model created by training the central model at the second endpoint using second local data, the second local data different from the first local data, the second local data local to the second endpoint;
access a third model from the third endpoint;
analyze one or more properties of the first endpoint, the second endpoint, and the third endpoint to select the first locally trained model and the second locally trained model for aggregation;
aggregate the first locally trained model and the second locally trained model to produce a new central model, the third model excluded from the new central model;
provide the new central model to the first endpoint; and
provide the new central model to the second endpoint.
Claims 1, 8, 16 (claim 1 exemplary)
At least one non-transitory computer readable storage medium comprising instructions to cause at least one processor circuit to at least:
distribute a central model to a first endpoint, a second endpoint, and a third endpoint;
access a first locally trained model, the first locally trained model created by training the central model at the first endpoint using first local data, the first local data local to the first endpoint;
access a second locally trained model, the second locally trained model created by training the central model at the second endpoint using second local data, the second local data different from the first local data, the second local data local to the second endpoint;
access a third model from the third endpoint;
aggregate the first locally trained model and the second locally trained model to produce a new central model, the third model excluded from the new central model as a result of the third endpoint being a non-trusted device;
provide the new central model to the first endpoint; and
provide the new central model to the second endpoint.
Claims 2, 9, 16 (claim 2 exemplary)
The at least one non-transitory computer-readable storage medium of claim 1, wherein the instructions cause one or more of the at least one processor circuit to access the first locally trained model without having access to the first local data and the instructions cause one or more of the at least one processor circuit to access the second locally trained model without having access to the second local data.
Claims 2, 9, 17 (claim 2 exemplary)
The at least one non-transitory computer readable
storage medium of claim 1, wherein the instructions cause one or more of the at least one processor circuit to access the first locally trained model without having access to the first local data and the instructions cause one or more of the at least one processor circuit to access the second locally trained model without having access to the second local data.
Claims 3, 10, 17 (claim 3 exemplary)
The at least one non-transitory computer-readable storage medium of claim 1, wherein the first endpoint is implemented using a first hardware configuration and the second endpoint is implemented using a second hardware configuration different from the first hardware configuration.
Claims 3, 10, 18 (claim 3 exemplary)
The at least one non-transitory computer readable
storage medium of claim 1, wherein the first endpoint is implemented using a first hardware configuration and the second endpoint is implemented using a second hardware configuration different from the first hardware configuration.
Claims 4, 11, 18 (claim 4 exemplary)
The at least one non-transitory computer-readable storage medium of claim 1, wherein the second locally trained model is formatted in an encrypted format.
Claims 4, 12 (claim 4 exemplary)
The at least one non-transitory computer readable
storage medium of claim 1, wherein the second locally trained model is formatted in an encrypted format.
Claims 5, 12, 19 (claim 5 exemplary)
The at least one non-transitory computer-readable storage medium of claim 4, wherein the at least one processor circuit cannot associate the second local data with the second endpoint.
Claims 5, 13 (claim 5 exemplary)
The at least one non-transitory computer readable storage medium of claim 4, wherein the at least one processor circuit cannot associate the second local data with the second endpoint.
Claims 6, 13, 20 (claim 6 exemplary)
The at least one non-transitory computer-readable storage medium of claim 1, wherein the instructions cause one or more of the at least one processor circuit to update the central model using at least a first portion of the first locally trained model and a second portion of the second locally trained model.
Claims 6, 14 (claim 6 exemplary)
The at least one non-transitory computer readable
storage medium of claim 1, wherein the instructions cause one or more of the at least one processor circuit to update the central model using at least a first portion of the first locally trained model and a second portion of the second locally
trained model.
Claims 7, 14 (claim 7 exemplary)
The at least one non-transitory computer-readable storage medium of claim 1, wherein the instructions to aggregate the first locally trained model and the second locally trained model are executed using a trusted execution environment of the at least one processor circuit.
Claims 7, 15 (claim 7 exemplary)
The at least one non-transitory computer readable
storage medium of claim 1, wherein the instructions to aggregate the first locally trained model and the second locally trained model are executed using a trusted execution environment of the at least one processor circuit.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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 1-6, 8-13 and 15-20 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Publication No. 2019/0012592 to Beser et al. (“Beser”) in view of U.S. Patent Publication No. 2018/0240011 to Tan et al. (“Tan”) and further in view of U.S. Patent No. 2020/0380340 to Blanchard et al. (“Blanchard”).
As to claim 1, Beser discloses at least one non-transitory computer-readable storage medium comprising instructions to cause at least one processor circuit (Beser: fig 1-3, [0008; 40]: non-transitory computer-readable medium 38 [0008; 40]) to at least:
distribute a central model to a first endpoint, a second endpoint, and a third endpoint (Beser: fig 1-3, [0011-41]: fig 3 … step 50 a central model is downloaded (distribute a central model to …) from a central server to first plurality of artificial neural networks (ANNs) (first endpoint(s)) and to a second plurality of artificial neural networks (ANNs) (second endpoint(s)) [0041] … ANN1 ANN2 ANN3 (third endpoint(s)) comprise one or more networks of related computers, cellphones, watches, mobile devices … may be located within single installation, multiple installations, single location, multiple locations etc [0030]);
access a first locally trained model, the first locally trained model created by training the central model at the first endpoint using first local data, the first local data local to the first endpoint (Beser: fig 1-3, [0011-41]: fig 3 … step 52 a first local model within first federation is computed based on the first local data as applied to central model [0041]);
access a second locally trained model, the second locally trained model created by training the central model at the second endpoint using second local data, the second local data different from the first local data, the second local data local to the second endpoint (Beser: fig 1-3, [0011-41]: fig 3 … as is a second local model within the second federation based on the second local data as applied to the central model at step 54 [0041] … securing communications between a central server and clusters of ANNs to update a central model of the central server from local datasets obtained at the edges of the ANNs (first endpoint(s) and second endpoint(s)) without directly exposing the central server to the local datasets … whereby unconnected federations receive benefit from external dataset trainings at other federations (using second local data, the second local data different from the first local data) [0042]);
access a third model from the third endpoint (Beser: fig 1-3, [0011-41]: … central server maintains a first second and third download connection D1 D2 D3 with the first second third ANN1 ANN2 ANN3 and in like fashion the first second third ANN1 ANN2 ANN3 maintain a first second and third upload connection D1 D2 D3 with central server [0031]).
Beser did not explicitly disclose analyze one or more properties of the first endpoint, the second endpoint, and the third endpoint to select the first locally trained model and the second locally trained model for aggregation.
Tan discloses analyze one or more properties of the first endpoint, the second endpoint, and the third endpoint to select the first locally trained model and the second locally trained model for aggregation (Tan: fig 1- 13; abstract, [0003-83]: ... In the data sampling techniques (analyze one or more properties of the first endpoint, the second endpoint, and the third endpoint ...), a subset (... to select the first locally trained model and the second locally trained model for aggregation) of the data obtained at the local sites is intelligently selected for transfer to the central site for use in training the central machine learning model and in model merging techniques, distributed local training occurs in each local site (analyze one or more properties of the first endpoint, the second endpoint, and the third endpoint ...) and copies of the local machine learning models are sent to a central site for aggregation of learning by merging the models (... to select the first locally trained model and the second locally trained model for aggregation)).
Beser and Tan are analogous art because they are from the same field of endeavor with respect to models.
Before the effective filing date, for AIA , it would have been obvious to a person of ordinary skill in the art to incorporate the strategies by Tan into the medium by Beser. The suggestion/motivation would have been to provide a central machine learning model that may be trained on various representations/transformations of data seen at local machine learning models, including sampled selections of data-label pairs etc (Tan: [0017]).
Beser did not explicitly disclose aggregate the first locally trained model and the second locally trained model to produce a new central model, the third model excluded from the new central model as Byzantine.
Blanchard discloses aggregate the first locally trained model and the second locally trained model to produce a new central model (Blanchard: fig 1-10, [0015-132]: fig 1 … a first computer “parameter server” and n worker computers similar to a general distributed system model … a portion f of the workers are possible “Byzantine” i.e. they may deliver erroneous and/or arbitrary results [0049] … deep learning solution deployed and trained over several computers (1ST 2ND 3RD ... n endpoints) and parameter vector is a vector comprising all the synaptic weights and the internal parameters of the deep learning model (1ST 2ND 3RD ... n locally trained models) … cost function is any measure of deviation between what deep learning model predicts and what the computers actually observe [0058] … the present method operates only on a subset (i.e. exclude certain models) of the received estimate vectors (aggregate the first locally trained model and the second locally trained model to produce a new central model) [0017]),
the third model excluded from the new central model as Byzantine (Blanchard: fig 1-10, [0015-132]: fig 1 … determining the updated parameter vector precludes (excludes) the estimate vectors which have a distance greater than a predefined maximum distance to the other estimate vectors (see with [0049;58;17] above - the third model excluded from the new central model as Byzantine) … the first computer does not aggregate all estimate vectors received from worker computers but disregards (excludes) vectors that are too far away from the other vectors, sinch such outliers are erroneous with high likelihood (see with [0049;58;17] above - the third model excluded from the new central model as Byzantine) [0018]).
Beser, Tan and Blanchard are analogous art because they are from the same field of endeavor with respect to models.
Before the effective filing date, for AIA , it would have been obvious to a person of ordinary skill in the art to incorporate the strategies by Blanchard into the medium by Beser and Tan. The suggestion/motivation would have been to provide a method different from an averaging approach that takes into account all vectors, even the erroneous ones (Blanchard: [0017]) and provide a distributed machine learning implementation that is both fault tolerant i.e. delivers correct results even in the presence of arbitrarily erroneous workers (see with [0049] – Byzantine) and efficient i.e. less computationally intensive (Blanchard: [0014]).
Beser, Tan and Blanchard further disclose provide the new central model to the first endpoint; and provide the new central model to the second endpoint (Beser: fig 1-3, [0011-41]: fig 3 … step 68 an updated central model from the central server is downloaded to (provide the new central model ...) the first plurality of artificial neural networks (ANNs) (first endpoint(s)) and to a second plurality of artificial neural networks (ANNs) (second endpoint(s)) [0041]).
Same motivation applies as mentioned above to make the proposed modification.
As to claim 2, Beser, Tan and Blanchard disclose wherein the instructions cause one or more of the at least one processor circuit to access the first locally trained model without having access to the first local data and the instructions cause one or more of the at least one processor circuit to access the second locally trained model without having access to the second local data (Beser: fig 1-3, [0011-41]: fig 3 … securing communications between a central server and clusters of ANNs to update a central model of the central server from local datasets obtained at the edges of the ANNs (first endpoint(s) and second endpoint(s)) without directly exposing the central server to the local datasets … whereby unconnected federations receive benefit from external dataset trainings at other federations [0042]).
For motivation, see rejection of claim 1.
As to claim 3, Beser, Tan and Blanchard disclose wherein the first endpoint is implemented using a first hardware configuration and the second endpoint is implemented using a second hardware configuration different from the first hardware configuration (Beser: fig 1-3, [0011-41]: ANNs and/or central server comprise one or more controllers comprising processors … configured to implement one or more ASIC, DSP, FPGA or various combinations for performing various functions [0039] … federated ANNs control their interaction with the central server and vice versa using an export schema to specify what information flows from an ANN and/or central server as well as import schema to specify what information flows into an ANN and/or the central server [0019]).
For motivation, see rejection of claim 1.
As to claim 4, Beser, Tan and Blanchard disclose wherein the second locally trained model is formatted in an encrypted format (Beser: fig 1-3, [0011-41]: fig 1-2 … when update from first local model is available, first ANN uploads the first update to the central server using authentication and encryption … when update from second local model is available, second ANN uploads the second update to the central server using authentication and encryption [0036] … messages sent between central server and ANNs are authenticated and encrypted [0025]).
For motivation, see rejection of claim 1.
As to claim 5, Beser, Tan and Blanchard disclose wherein the at least one processor circuit cannot associate the second local data with the second endpoint (Tan: fig 6-13, [0057-94]: … model merging techniques only transmit model parameters and not the actual data or gradient values from each local node to the central site for performing global learning … this enhances privacy since the raw data does not leave the local sites and model parameters sent provide very limited information about the ensemble of data at the local sites [0069; 86]).
For motivation, see rejection of claim 1.
As to claim 6, see similar rejection to claim 5 where the medium is taught by the medium.
As to claims 8-13, see similar rejection to claims 1-6, respectively where the server is taught by the medium.
As to claims 15-20, see similar rejection to claims 1-6, respectively where the server is taught by the medium.
Claims 7 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Publication No. 2019/0012592 to Beser et al. (“Beser”) in view of U.S. Patent Publication No. 2018/0240011 to Tan et al. (“Tan”), U.S. Patent No. 2020/0380340 to Blanchard et al. (“Blanchard”) and further in view of U.S. Patent Publication No. 2021/0192360 A1 to Bitauld et al. (“Bitauld”).
As to claim 7, Beser, Tan and Blanchard disclose the medium of claim 1.
For motivation, see rejection of claim 1.
Beser did not explicitly disclose wherein the instructions to aggregate the first locally trained model and the second locally trained model are executed using a trusted execution environment of the at least one processor circuit.
Bitauld discloses wherein the instructions to aggregate the first locally trained model and the second locally trained model are executed using a trusted execution environment of the at least one processor circuit (Bitauld: fig 1-5 [0029-74]: fig 5 block 530 running, in trusted execution environment, training process configured to obtain parameters of neural network, using training data … fig 1-2 … process of training neural network takes place in TEE (trusted execution environment)108 [0051] … a so-called ‘student-teacher’ approach to train more private network(s) (first second … n locally trained model(s)) and these systems work by training ensembles of ‘teachers’ on subsets of the private data, after the ensemble is trained a ‘student’ is trained to predict the aggregate output of the ‘teachers’ on publicly available and potentially unlabeled data and, in this way, the ‘student’ network can never be reverse engineered to reveal original private data [0055] … the trained neural network remains in the TEE or another TEE [0056]).
Beser, Tan, Blanchard and Bitauld are analogous art because they are from the same field of endeavor with respect to neural networks.
Before the effective filing date, for AIA , it would have been obvious to a person of ordinary skill in the art to incorporate the strategies by Bitauld into the medium by Beser, Tan and Blanchard. The suggestion/motivation would have been to provide training neural network(s) in a TEE to provide the advantage of training data is concealed from outside parties during transmission, storage and processing necessary during neural network training processes (Bitauld: [0044]).
As to claim 14, see similar rejection to claim 7 where the server is taught by the medium.
Conclusion
The following prior art made of record and not relied upon is considered pertinent to applicant’s disclosure.
A) US 20250278474 – Salonidis
An intermediate global lower-level machine learning model is generated by executing federated learning model training using data from an identified first set of lower-level non-Byzantine client computers which are characterized as being non-Byzantine for the lower-level system. A global lower-level machine learning model is generated by executing the federated learning model training using data from an identified second lower-level set of non-Byzantine client computers which are characterized as being non-Byzantine for the lower-level system. An intermediate global upper-level machine learning model is generated by executing the federated learning model training using data from an identified first upper-level set of non-Byzantine client computers which are characterized as being non-Byzantine for an upper-level system. A global upper-level machine learning model is generated by executing the federated learning model training using data from a second upper-level set of client computers which are characterized as being non-Byzantine for the upper-level system.
B) US 20220292387 – Zhou
Embodiments of the present disclosure include a federated learning method by a federated learning aggregator. The method may comprise creating a log of previously provided gradients from a plurality of workers, receiving updated gradients from the plurality of workers, calculating a vulnerability weight for each layer of a global machine learning model using the updated gradients, calculating an aggregated gradient using the vulnerability weight and the updated gradients, and updating the global machine learning model using the aggregated gradient. Some embodiments may also determine whether a Byzantine attack is occurring based upon the calculated aggregated gradient.
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 JUNE SISON whose telephone number is (571)270-5693. The examiner can normally be reached 9:00 am - 5:00 pm.
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, Emmanuel Moise can be reached at 571-272-3865. 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.
1
/JUNE SISON/Primary Examiner, Art Unit 2455
Prosecution Insights