DETAILED ACTION
This action is responsive to RCE filed on June 30, 2026.
Claims 29-39, 44-45 and 48 have been amended.
Claims 29-39 and 41-49 are pending and are presented to examination.
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on June 30, 2026 has been entered.
Examiner Notes
Examiner cites particular columns, paragraphs, figures and line numbers in the references as applied to the claims below for the convenience of the applicant. Although the specified citations are representative of the teachings in the art and are applied to the specific limitations within the individual claim, other passages and figures may apply as well. It is respectfully requested that, in preparing responses, the applicant fully consider the references in their entirety as potentially teaching all or part of the claimed invention, as well as the context of the passage as taught by the prior art or disclosed by the examiner.
Response to Arguments
Applicant’s arguments filed June 30, 2026 with respect to the rejection of independent claims 29, 39, and 48 under 35 U.S.C. 103 have been fully considered but are not persuasive.
Applicant argues (Applicant’s Remarks) that, as shown in FIG. 1 of Lupesko, the adaptive model engine 109, the model 107, and the model variant 108 are located at the edge device 101, and that the generation of the model variant (e.g., at 809) is likewise performed at the edge device, such that Bernat in view of Lupesko purportedly does not teach the amended limitations directed to sending to, and receiving a model configuration from, at least one second device.
The examiner respectfully disagrees. Lupesko is not limited to adaptation performed locally at the edge device. In addition to the adaptive model engine 109 that may execute on the edge device, Lupesko expressly discloses an adaptive model service 113 that executes on a separate web services provider 111 (see FIG. 7) — i.e., a second device. As shown in FIGs. 9-10, the edge device (first device) issues a call to the adaptive model service that provides a model identifier, the desired performance, and the current device characteristics, and the service, in response, selects or generates and sends a model variant or profile back to the requesting device. See Lupesko, col. 6 lines 38-45 (FIG. 9, step 901), “an API call from an edge device is received. The API call provides the desired performance of the model, a model identifier (or the model itself), and current device characteristics”; col. 7 lines 8-11 (FIG. 9, step 909) (the selected model, generated model variant, or profile “is sent to the requesting device”); and Abstract, “sending the generated or selected model variant or profile to the edge device to use in inference”. Lupesko further teaches that even the edge-device engine’s selection or generation of a variant “includes a call to an adaptive model service which provides a model variant or a profile” (col. 6 lines 23-30; see also col. 5 line 62 – col. 6 line 9). Applicant’s reliance on FIG. 1 and the local engine 109 therefore does not address FIG. 7 (adaptive model service 113) or the methods of FIGs. 9-10 performed by the adaptive model service on the web services provider, which supply the claimed sending to, and receiving of the first and second model configurations from, at least one second device.
Applicant further argues that the claims have been amended to recite that the request for a model update is associated with the second resource constraint, and that this feature has not been established.
The examiner respectfully disagrees. Lupesko discloses that, after the model variant is used for inference, “at a later point, current device characteristics are received” (col. 6 line 31-35, FIG. 8, step 813), and that when the changed characteristics indicate that the model no longer meets the desired performance, a further call is made to the adaptive model service (col. 5 line 34-43; col. 5 line 62 – col 6 line 9). That further call carries the current, changed device characteristics, which the adaptive model service uses to select or generate the updated model configuration (col. 6 line 38-45, FIG. 9, step 901). The current, changed device characteristics sent with the request for the model update thus constitute the indication of a second resource constraint that is associated with the model update, as recited. The limitation is accordingly taught by Bernat in view of Lupesko as set forth in the rejection below.
For at least the foregoing reasons, Applicant’s arguments are not persuasive, and claims 29-39 and 41-49 remain rejected for the reasons set forth in the rejections above.
Claim Objections
Claims 39 and 41-47 are objected to because of the following informalities: Claim 39 recites the limitation ”based on processing the first data, sending to the at least one second device a request for a model update and an indication of a second resource constraint associated with the first device, wherein the second resource constraint is associated with the model update;” in lines 9-11. Please add a comma “,” after “device”. Appropriate correction is required.
Dependent claims 41-47 do not overcome the deficiency of the base claim and, therefore, are objected for the same reasons as the base claim.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 31 and 41 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 31 (and similar for claim 41) recites "wherein the second model configuration comprises the second model.". There is insufficient antecedent basis for this limitation in the claim.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 29-36, 38-39, 41-45 and 47-49 are rejected under 35 U.S.C. 103 as being unpatentable over Guim Bernat et al. (US Pub. No. 2019/0044831, hereinafter Bernat – previously presented) in view of Lupesko et al. (US Pat. No. 11,423,283, hereinafter Lupesko – previously presented).
With respect to claim 29 (Currently Amended), Bernat teaches a first device comprising: a transceiver (Bernat, Figures 5-6 (mesh transceiver 662 and wireless network transceiver 666); paragraph [0029], “each IoT device 104 may include other transceivers for communications using additional protocols and frequencies”. Bernat discloses an IoT device (e.g., IoT device 650 of Fig. 6) — a first device — that includes one or more transceivers (662, 666) enabling wireless communication with other devices.) and a processor configured to: (Bernat, Figure 6 (processor 652); paragraphs [0080], [0097], “code to direct the processor 652 to perform electronic operations in the IoT device 650”. The IoT device 650 includes a processor 652 configured to execute instructions to perform the recited operations.) send, via the transceiver to at least one second device, a model identifier identifying a model and an indication of a first resource constraint associated with the first device (Bernat, paragraph [0038], “The edge clients send inference requests to the edge network platform ... It allows each client to specify a Model ID and an optional requirement such as deadline, performance or cost”; see also Fig. 3 (model ID 310), paragraph [0040]; Abstract. Bernat’s edge client (first device) sends inference requests, via its transceiver, to the edge network platform / gateway (at least one second device). Each request specifies a “Model ID” — a model identifier identifying a model — together with an optional requirement such as a deadline, performance, or cost target — an indication of a first resource constraint associated with the requesting first device.)
Bernat is silent to disclose, however, in an analogous art, Lupesko teaches: receive, via the transceiver from the at least one second device in response to the model identifier and the indication of the first resource constraint, a first model configuration of the identified model, wherein the first model configuration is configured to operate within the first resource constraint associated with the first device (Lupesko, col. 6 lines 38-45; FIG. 9 (steps 901), “an API call from an edge device is received. The API call provides the desired performance of the model, a model identifier ..., and current device characteristics”; col. 7 lines 8-11 (FIG. 9, step 909), “the selected model, generated model variant, or selected profile is sent to the requesting device”; Abstract; col. 2 lines 45-57. In Lupesko, the edge device (first device) issues an API call to the adaptive model service 113 executing on the web services provider 111 — the second device to which the request is directed — providing a model identifier, the desired performance, and the current device characteristics. In response, the service selects or generates a model variant or profile that meets the desired performance given those characteristics and sends it to the requesting device. The returned variant/profile is a first model configuration of the identified model configured to operate within the first resource constraint. In the combination, the second device to which Bernat’s first device sends the model identifier and resource constraint responds, as taught by Lupesko, by returning such an adapted first model configuration to the first device for local execution.) process first data using the first model configuration (Lupesko, FIG. 8 (step 813), col. 6 lines 31-35, “the selected variant model or dynamically generated model variant ... is used for inference”. The edge device runs the received model variant on its input data to perform inference.) based on processing the first data, send, via the transceiver to the at least one second device, a request for a model update and an indication of a second resource constraint associated with the first device, wherein the second resource constraint is associated with the model update (Lupesko, col. 5 lines 34-43 (FIG. 8, step 800), “when operating parameters of the edge device change, or what is acceptable performance changes, the adaptive model engine is called by software monitoring usage of the model”; col. 6 lines 31-35 (step 813), “at a later point, current device characteristics are received”; col. 5 line 62 – col. 6 line 9 (step 809), “a call to an adaptive model service is made which, in turn, provides a profile to use”; col. 6 lines 38-45 (FIG. 9, step 901). After the model variant is used for inference, current device characteristics are received at a later point (col. 6 lines 31-35); when the changed characteristics indicate the model no longer meets the desired performance, the edge device issues a further call to the adaptive model service (second device) — a request for a model update — carrying the current, changed device characteristics, which constitute an indication of a second resource constraint that is associated with, i.e., accompanies and defines, the requested model update.) receive, via the transceiver from the at least one second device in response to the request for the model update and the indication of the second resource constraint, a second model configuration, wherein the second model configuration is configured to operate within the second resource constraint associated with the first device (Lupesko, col. 5 line 62 – col. 6 line 9 (FIG. 8, step 809), “a profile of a model variant that will meet the characteristics is selected ... and used to generate a model variant”; col. 7 lines 8-11 (FIG. 9, step 909); Abstract. In response to the update request and the updated device characteristics, the adaptive model service selects/generates and returns a further model variant/profile — a second model configuration — configured to operate within the second resource constraint.) and process second data using the second model configuration (Lupesko, col. 6 lines 31-35 (FIG. 8, step 813), “the selected variant model or dynamically generated model variant ... is used for inference”. The edge device runs the updated model variant on subsequent input data.)
It would have been obvious to one of ordinary skill in the art at the time the invention was made before the effective filing date of the claimed invention to modify the first device of Bernat — which sends, to a second device, a model identifier and an indication of a resource constraint associated with the device — such that the second device generates or selects and returns to the first device a model configuration that operates within that resource constraint for local execution, and such that the first device iteratively requests and receives an updated model configuration as the resource constraint changes, as taught by Lupesko. One of ordinary skill would have been motivated to make this combination in order to allow the requesting device to continue operating within its changing computational-resource limitations — rather than being shut down or suffering degraded performance when the available resources fall below the model’s requirements — thereby improving data processing and providing a mechanism for dynamic machine-learning model selection and adaptation (Lupesko, col. 1 line 53 – col. 2 line 9; Abstract).
With respect to claim 30 (Currently Amended), Bernat teaches wherein the first device comprises a smartphone (Bernat, paragraph [0019], “an IoT device may be a smart phone, laptop, tablet, or PC, or other larger device”. Bernat expressly discloses that the IoT device (first device) may be a smart phone.)
With respect to claim 31 (Currently Amended), Bernat is silent to disclose, however, in an analogous art, Lupesko teaches wherein the second model configuration comprises the second model (Lupesko, Abstract, “a model variant or a model variant profile”; col. 7 lines 8-11 (FIG. 9, step 909); col. 2 lines 37-44, “a ML model 107 (or a variant model 108)”. Lupesko discloses that the configuration returned to the edge device may be a complete model variant (model′ 108) rather than merely a profile, such that the second model configuration comprises the second model itself.)
It would have been obvious to one of ordinary skill in the art at the time the invention was made before the effective filing date of the claimed invention to provide, as the returned second model configuration, a complete model variant as taught by Lupesko, for the same reasons and with the same motivation set forth with respect to claim 29 above, namely to enable the first device to operate within its computational-resource limitations through dynamic model selection and adaptation (Lupesko, col. 1 line 53 – col. 2 line 9; Abstract). With respect to claim 32 (Currently Amended), Bernat is silent to disclose, however, in an analogous art, Lupesko teaches wherein the second model configuration comprises at least one parameter update to the first model configuration (Lupesko, col. 7 lines 43-50, “One or more parameters of the received model are changed ... quantizing (such as going from 32-bit floating point to 8-bit integer), layer pruning, changing or fusing operators, and/or unrolling a neural network”. Lupesko teaches that the returned configuration changes one or more parameters of the model (e.g., quantization, pruning, operator changes), i.e., at least one parameter update to the first model configuration.)
It would have been obvious to one of ordinary skill in the art at the time the invention was made before the effective filing date of the claimed invention to implement the second model configuration as at least one parameter update to the first model configuration as taught by Lupesko, for the same reasons and with the same motivation set forth with respect to claim 29 above (Lupesko, col. 1 line 53 – col. 2 line 9; Abstract). With respect to claim 33 (Currently Amended), Bernat is silent to disclose, however, in an analogous art, Lupesko teaches wherein the second model configuration is configured to operate within the first resource constraint and within the second resource constraint. (Lupesko, col. 2 line 58 – col. 3 line 2, “the edge device 101 stores a variant of the model 107 (model variant 108 ...) or a profile to be used to change the model 107”; Abstract. Lupesko generates each variant/profile to satisfy the desired performance under the device’s characteristics, such that the second model configuration operates within both the initial (first) and the changed (second) resource constraints.)
It would have been obvious to one of ordinary skill in the art at the time the invention was made before the effective filing date of the claimed invention to configure the second model configuration to operate within both the first and second resource constraints as taught by Lupesko, for the same reasons and with the same motivation set forth with respect to claim 29 above (Lupesko, col. 1 line 53 – col. 2 line 9; Abstract). With respect to claim 34 (Currently Amended), Bernat is silent to disclose, however, in an analogous art, Lupesko teaches wherein at least one of the first resource constraint or the second resource constraint comprises at least one of a limit on computational resources associated with the first device, or an accuracy constraint (Lupesko, col. 2 lines 45-57, “FLOPS, GPU RAM, CPU RAM, CPU speed, power, network capabilities ..., memory, etc.; ... desired objectives of ML execution (such as throughput, power usage, and accuracy)”. Lupesko’s constraints include limits on computational resources of the device (FLOPS, RAM, CPU, power, memory) and an accuracy objective.)
It would have been obvious to one of ordinary skill in the art at the time the invention was made before the effective filing date of the claimed invention to define the resource constraint as a limit on computational resources or an accuracy constraint as taught by Lupesko, for the same reasons and with the same motivation set forth with respect to claim 29 above (Lupesko, col. 1 line 53 – col. 2 line 9; Abstract). With respect to claim 35 (Currently Amended), Bernat is silent to disclose, however, in an analogous art, Lupesko teaches wherein at least one of the first resource constraint or the second resource constraint comprises a resource availability constraint at the first device (Lupesko, col. 5 lines 50-53 (FIG. 8, step 803), “how much memory is available, what power source is being used (battery or AC), processor types, etc. are received”; col. 45-27. Lupesko’s current device characteristics (available memory, available power) are a resource availability constraint at the first device.)
It would have been obvious to one of ordinary skill in the art at the time the invention was made before the effective filing date of the claimed invention to define the resource constraint as a resource availability constraint at the first device as taught by Lupesko, for the same reasons and with the same motivation set forth with respect to claim 29 above (Lupesko, col. 1 line 53 – col. 2 line 9; Abstract). With respect to claim 36 (Currently Amended), Bernat teaches wherein the second resource constraint is based on the first device moving to an edge computing node close to the first device (Bernat, paragraph [0013], “a car that is about to leave an area of coverage in the next fraction of a second may need an inference ... within a few hundred milliseconds”; see also paragraphs [0029]-[0030]. Bernat describes an edge/fog topology in which a first device (e.g., a vehicle) moves relative to, and requests inference from, nearby edge-cloud platforms/nodes, such that the applicable resource constraint (e.g., available resources or required response time) is based on the first device’s movement/proximity to an edge computing node close to the device.) With respect to claim 38 (Currently Amended), Bernat is silent to disclose, however, in an analogous art, Lupesko teaches wherein the first model configuration and the second model configuration comprise a neural network (Lupesko, col. 3 lines 26-35 (FIG. 2), “a deep learning model in the form of an artificial neural network”; col. 2 line 37-44 (ML model 107 and variant 108). Lupesko’s model 107 and its variants are deep-learning artificial neural networks, such that the first and second model configurations comprise a neural network.)
It would have been obvious to one of ordinary skill in the art at the time the invention was made before the effective filing date of the claimed invention to implement the first and second model configurations as a neural network as taught by Lupesko, for the same reasons and with the same motivation set forth with respect to claim 29 above (Lupesko, col. 1 line 53 – col. 2 line 9; Abstract).
With respect to claim 39 (currently amended), the claim recites limitations similar to those of claim 29, differing only in that it is directed to a method performed by a first device rather than to the first device itself, and is rejected under the same rationale set forth for claim 29 above.
With respect to claim 41 (previously presented), the claim recites limitations similar to claim 31 and is rejected for the same reasons set forth for claim 31 above.
With respect to claim 42 (previously presented), the claim recites limitations similar to claim 32 and is rejected for the same reasons set forth for claim 32 above.
With respect to claim 43 (previously presented), the claim recites limitations similar to claim 33 and is rejected for the same reasons set forth for claim 33 above.
With respect to claim 44 (currently amended), the claim recites limitations similar to claim 35 and is rejected for the same reasons set forth for claim 35 above.
With respect to claim 45 (currently amended), the claim recites limitations similar to claim 36 and is rejected for the same reasons set forth for claim 36 above.
With respect to claim 47 (previously presented), the claim recites limitations similar to claim 38 and is rejected for the same reasons set forth for claim 38 above.
With respect to claim 48 (currently amended), the claim recites limitations similar to those of claim 29, differing only in that it is directed to at least one non-transitory computer-readable storage medium having executable instructions stored thereon that, when executed by a processor, cause the processor to perform the recited operations, and is rejected under the same rationale set forth for claim 29 above. The first-device and second-device recitations of claim 48 correspond to the first-device and second-device operations of claim 29. Bernat further teaches such a medium (Bernat, paragraph [0097], “the instructions 682 ... embodied as a non-transitory, machine readable medium 660 including code to direct the processor 652 to perform electronic operations”).
With respect to claim 49 (previously presented), the claim recites limitations similar to claim 35 and is rejected for the same reasons set forth for claim 35 above.
Claims 37 and 46 are rejected under 35 U.S.C. 103 as being unpatentable over Guim Bernat et al. (US Pub. No. 2019/0044831) in view of Lupesko et al. (US Pat. No. 11,423,283) and further in view of Song et al. (US Pub. No. 2017/0025119, hereinafter Song). With respect to claim 37 (Currently Amended), Bernat in view of Lupesko is silent to disclose, however, in an analogous art, Song teaches wherein at least one of the first model configuration or the second model configuration is configured to process at least one of the first data or the second data based on a windowing function (Song, paragraph [0058], “The preprocessor 110 converts an audio signal to be recognized into audio frames, and extracts the audio frames into windows ... sequentially extracts the audio frames by dividing the audio frames into successive windows”; see also paragraphs [0059], [0065]-[0066], “Each time a window is extracted by the preprocessor 110, the score calculator 120 inputs frames included in the extracted window to an acoustic model”. Song discloses a deep-neural-network-based model whose input data (an audio signal) is processed based on a windowing function: a preprocessor divides the incoming frames into successive windows (W1, W2, ...), and the model processes the data window-by-window. Accordingly, at least one model configuration is configured to process the data based on a windowing function.) It would have been obvious to one of ordinary skill in the art at the time the invention was made before the effective filing date of the claimed invention to further modify the model configuration of Bernat in view of Lupesko — which operates within a resource constraint of the first device — so as to process the input data based on a windowing function as taught by Song, i.e., by dividing the input data into successive windows input to the neural-network model. One of ordinary skill would have been motivated to make this combination because Song teaches that the window size may be determined, and dynamically adjusted, based on the computing performance capability of the applied device (Song, paragraphs [0060], [0063]-[0064]); windowed processing therefore directly complements, and provides a further tunable mechanism for, the resource-constraint-based model adaptation of Bernat and Lupesko, matching the model’s computational cost to the device’s available resources while enabling incremental, real-time processing on resource-limited devices (Song, paragraph [0006]).
With respect to claim 46 (previously presented), the claim recites limitations similar to claim 37 and is rejected for the same reasons set forth for claim 37 above.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ANIBAL RIVERACRUZ whose telephone number is (571)270-1200. The examiner can normally be reached Monday-Friday 9:30 AM-6:00 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, Hyung S Sough can be reached at 5712726799. 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.
/ANIBAL RIVERACRUZ/Primary Examiner, Art Unit 2192
Prosecution Insights