DETAILED ACTION
This action is in response to the reply filed 23 June 2026.
Claims 1–20 are pending. Claims 1, 8, and 15 are independent.
Claims 1–20 are rejected.
Notice of Pre-AIA or AIA Status
The present application, filed on or after 16 March 2013, is being examined under the first inventor to file provisions of the AIA .
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.
Response to Arguments
The objection to the abstract is withdrawn in light of the new copy of the abstract.
Applicant's arguments, see remarks, filed 23 June 2026, with respect to the rejection(s) of claim(s) 1–20 under §§ 102 and 103 have been fully considered and are persuasive. Therefore, the rejections have been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of Hunt et al.
Claim Rejections—35 U.S.C. § 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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 C.F.R. § 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. § 102(b)(2)(C) for any potential 35 U.S.C. § 102(a)(2) prior art against the later invention.
Claims 1, 3, and 4 are rejected under 35 U.S.C. § 103 as being unpatentable over Klein et al. (US 2022/0239758 A1) [hereinafter Klein] in view of Hunt et al. (US 2018/0211153 A1) [hereinafter Hunt].
Regarding independent claim 1, Klein teaches [a] method for performing machine learning and data analytics, the method comprising: A method for routing machine learning requests (Klein, abstract). The machine learning may be for data analytics (Klein, ¶ 80). determining a quantum of historical data associated with a machine learning model; The machine learning model may use historical data, e.g., historical malware signatures, to make predictions (Klein, ¶ 94). Machine learning capabilities of network devices are compared to a machine learning request, including memory capacity/availability, storage capacity, data transmission rate, etc. [capabilities related to amounts/quanta of data] (Klein, ¶¶ 49–53). The machine learning request is routed based on the processing load, including memory consumption (Klein, ¶ 135). determining an order associated with the machine learning model; The machine learning capabilities include the ability to process neural networks of a particular size or topology, models of a particular type, hyperparameters, etc. [complexity/order of the models] (Klein, ¶¶ 48, 50, 53, 159, 163). determining whether a latency associated with the machine learning model is critical; The machine learning request may have a QoS [quality of service] requirement, including a latency requirement (Klein, ¶ 135). selecting a server from a plurality of servers based at least in part on the quantum of historical data, the order, and the latency; The machine learning request is routed to a network device based on a comparison of the request and the capabilities and characteristics of the available devices (Klein, ¶¶ 49, 159, 167, 168). […]
Klein teaches training the machine learning model using the server (Klein, ¶¶ 79, 80) but does not expressly teach reconstructing a model from a weight matrix. However, Hunt teaches: receiving from the server, one or more structural details of the machine learning model and a corresponding weight matrix representation of the machine learning model; A plurality of neural network models [machine learning models] are stored as weight matrices and tensors (Hunt, ¶ 94). The storage media may be part of a distributed computing system and accessed using a communication interface (Hunt, ¶ 26). reconstructing the machine learning model from the one or more structural details of the machine learning model and the corresponding weight matrix representation; and A plurality of neural network models [machine learning models] are initialized by using weight matrices and tensors learned from a training step to reconstruct the models (Hunt, ¶ 94). executing the reconstructed machine learning model. The models initialized using the weight matrices are executed (Hunt, ¶ 95).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein with those of Hunt. Doing so would have been a matter of simple substitution of one known element [the model storage method of Klein] for another [the model storage method of Hunt] to obtain predictable results [a method of selecting a server based on a latency of a model, wherein the model is stored as a weight matrix].
Regarding dependent claim 3, the rejection of claim 1 is incorporated and Klein/Hunt further teaches: wherein the quantum of historical data associated with the machine learning model comprises one of a low quantum of historical data, a medium quantum of historical data, or a high quantum of historical data. The model may be a large-scale model with distributed input data [a high quantum of data] and processed by multiple network devices (Klein, ¶ 57). Machine learning capabilities of network devices are compared to a machine learning request, including memory capacity/availability, storage capacity, data transmission rate, etc. [capabilities related to amounts/quanta of data] (Klein, ¶¶ 49–53). [The requests are matched to servers based in part on their memory/storage capacity, therefore some requests require a lower capacity and some require a higher capacity.]
Regarding dependent claim 4, the rejection of claim 1 is incorporated and Klein further teaches: wherein the order associated with the machine learning model comprises one of a reduced order, a moderately high order, or a complex order. The machine learning processing capabilities include a size of neural network, number of layers or neurons, etc. [therefore machine learning requests are matched based on these aspects of model order/complexity, and the model orders may be higher or lower] (Klein, ¶ 48).
Claim 2 is rejected under 35 U.S.C. § 103 as being unpatentable over Klein et al. (US 2022/0239758 A1) [hereinafter Klein] in view of Hunt et al. (US 2018/0211153 A1) [hereinafter Hunt], further in view of Hayashi et al. (US 2023/0191474 A1) [hereinafter Hayashi].
Regarding dependent claim 2, the rejection of claim 1 is incorporated and Klein/Hunt further teaches: wherein the server is one of a cloud server or […], wherein reconstructing the machine learning model is performed by a model serializer of an embedded edge server, and further wherein the reconstructed machine learning model is executed by a machine learning model training and execution module of the embedded edge server. The network devices/nodes may be servers, including cloud servers and/or edge servers (Klein, ¶¶ 38–40, 52). The network devices include an integrated machine learning core having a pre-processing module, which includes a training component [training module] and local inference module [execution module] (Klein, ¶¶ 77–81).
Klein/Hunt teaches performing machine learning functions on edge servers and cloud servers, but does not expressly teach FOG servers. However, Hayashi teaches: a Free and Open-Source Ghost (FOG) server A machine learning device predicts a quality of output of a connected device; the machine learning device may be a FOG server (Hayashi, ¶ 82).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein/Hunt with those of Hayashi. Doing so would have been a matter of applying a known technique [hosting a ML model on a FOG server] to a known method ready for improvement [the ML model hosting of Klein] to yield predictable results [a method of routing ML models to cloud, edge, and FOG servers].
Claims 5–8, 10–15, and 17–20 are rejected under 35 U.S.C. § 103 as being unpatentable over Klein et al. (US 2022/0239758 A1) [hereinafter Klein] in view of Hunt et al. (US 2018/0211153 A1) [hereinafter Hunt], further in view of Zyglowicz et al. (US 2014/0365191 A1) [hereinafter Zyglowicz].
Regarding dependent claim 5, the rejection of claim 1 is incorporated. Klein/Hunt teaches routing machine learning requests to servers based on aspects of the requests compared to the servers, but does not expressly teach performing machine learning related to power transformers. However, Zyglowicz teaches: further comprising: calculating a first rate of change associated with an event in a power transformer system, wherein the event occurs during a first time period; A rate of change of dissolved hydrogen associated with insulation degradation of a power transformer, based on data for a prediction period (Zyglowicz, ¶ 66). calculating, using the reconstructed machine learning model, a first risk associated with the power transformer system based at least in part on the first rate of change; A health profile is generated based on the prediction period, including a risk, e.g., a risk of a fire event if a maintenance action correcting the degraded insulation is not performed within a timeframe (Zyglowicz, ¶¶ 74–76). The health profile is generated using a machine learning algorithm generated based on historical data (Zyglowicz, ¶ 31). calculating a second rate of change associated with the power transformer system, wherein the second rate of change is determined over a second time period; The industrial asset [transformer] may be monitored for multiple assessment/prediction periods (Zyglowicz, ¶ 61). calculating, using the reconstructed machine learning model, a second risk associated with the power transformer system based at least in part on the second rate of change; and The multiple assessment/prediction periods may be used to generate multiple health profiles (Zyglowicz, ¶ 61). determining a condition deterioration associated with the power transformer system based at least in part on the first risk and the second risk. The health profile may indicate, e.g., degradation [deterioration] of the insulation of a transformer (Zyglowicz, ¶¶ 65–67).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein/Hunt with those of Zyglowicz. Doing so would have been a matter of applying a known technique [routing ML requests to optimal servers] to a known method ready for improvement [the ML method for power transformers] to yield predictable results [a ML method for power transformers wherein the ML is routed to optimal servers based on comparing the models to the servers].
Regarding dependent claim 6, the rejection of claim 5 is incorporated and Klein/Hunt/Zyglowicz further teaches: further comprising: outputting a health status indicative of the condition deterioration to an operator. A health profile [health status indicator] is generated, including a recommended maintenance plan for the user to follow (Zyglowicz, ¶ 61). The health profile includes an indication of, e.g., degraded insulation (Zyglowicz, ¶ 65).
Regarding dependent claim 7, the rejection of claim 5 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the first risk and the second risk are calculated via a dissolved gas analyzer analytics engine, and wherein the dissolved gas analyzer analytics engine utilizes the reconstructed machine learning model, via historical data associated with the power transformer system, and real-time data associated with the power transformer system. The data used by the machine learning algorithm may include dissolved gas readings (Zyglowicz, ¶¶ 46, 52, 53, 57, 62–65). The data used may be past [historical] data, present [real-time] data, or estimated future data (Zyglowicz, ¶ 52).
Regarding independent claim 8, Klein teaches [a] method for performing machine learning and data analytics, the method comprising: A method for routing machine learning requests (Klein, abstract). The machine learning may be for data analytics (Klein, ¶ 80). determining a quantum of historical data associated with a machine learning model; The machine learning model may use historical data, e.g., historical malware signatures, to make predictions (Klein, ¶ 94). Machine learning capabilities of network devices are compared to a machine learning request, including memory capacity/availability, storage capacity, data transmission rate, etc. [capabilities related to amounts/quanta of data] (Klein, ¶¶ 49–53). The machine learning request is routed based on the processing load, including memory consumption (Klein, ¶ 135). determining an order associated with the machine learning model; The machine learning capabilities include the ability to process neural networks of a particular size or topology, models of a particular type, hyperparameters, etc. [complexity/order of the models] (Klein, ¶¶ 48, 50, 53, 159, 163). determining whether a latency associated with the machine learning model is critical; The machine learning request may have a QoS [quality of service] requirement, including a latency requirement (Klein, ¶ 135). selecting a server from a plurality of servers based at least in part on the quantum of historical data, the order, and the latency; The machine learning request is routed to a network device based on a comparison of the request and the capabilities and characteristics of the available devices (Klein, ¶¶ 49, 159, 167, 168). […]
Klein teaches training the machine learning model using the server (Klein, ¶¶ 79, 80) but does not expressly teach reconstructing a model from a weight matrix. However, Hunt teaches: receiving from the server, one or more structural details of the machine learning model and a corresponding weight matrix representation of the machine learning model; A plurality of neural network models [machine learning models] are stored as weight matrices and tensors (Hunt, ¶ 94). The storage media may be part of a distributed computing system and accessed using a communication interface (Hunt, ¶ 26). reconstructing the machine learning model from the one or more structural details of the machine learning model and the corresponding weight matrix representation; and A plurality of neural network models [machine learning models] are initialized by using weight matrices and tensors learned from a training step to reconstruct the models (Hunt, ¶ 94).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein with those of Hunt. Doing so would have been a matter of simple substitution of one known element [the model storage method of Klein] for another [the model storage method of Hunt] to obtain predictable results [a method of selecting a server based on a latency of a model, wherein the model is stored as a weight matrix].
Klein/Hunt teaches routing machine learning requests to servers based on aspects of the requests compared to the servers, but does not expressly teach performing machine learning related to power transformers. However, Zyglowicz teaches: determining, using the reconstructed machine learning model, a condition deterioration associated with a power transformer system. A health profile is generated based on a prediction period, including a risk, e.g., a risk of a fire event if a maintenance action correcting a degraded insulation of a power transformer, is not performed within a timeframe (Zyglowicz, ¶¶ 74–76). The health profile is generated using a machine learning algorithm generated based on historical data (Zyglowicz, ¶ 31).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein/Hunt with those of Zyglowicz. Doing so would have been a matter of applying a known technique [routing ML requests to optimal servers] to a known method ready for improvement [the ML method for power transformers] to yield predictable results [a ML method for power transformers wherein the ML is routed to optimal servers based on comparing the models to the servers].
Regarding dependent claim 10, the rejection of claim 8 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the quantum of historical data associated with the machine learning model comprises one of a low quantum of historical data, a medium quantum of historical data, or a high quantum of historical data. The model may be a large-scale model with distributed input data [a high quantum of data] and processed by multiple network devices (Klein, ¶ 57). Machine learning capabilities of network devices are compared to a machine learning request, including memory capacity/availability, storage capacity, data transmission rate, etc. [capabilities related to amounts/quanta of data] (Klein, ¶¶ 49–53). [The requests are matched to servers based in part on their memory/storage capacity, therefore some requests require a lower capacity and some require a higher capacity.]
Regarding dependent claim 11, the rejection of claim 8 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the order associated with the machine learning model comprises one of a reduced order, a moderately high order, or a complex order. The machine learning processing capabilities include a size of neural network, number of layers or neurons, etc. [therefore machine learning requests are matched based on these aspects of model order/complexity, and the model orders may be higher or lower] (Klein, ¶ 48).
Regarding dependent claim 12, the rejection of claim 8 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the determination, using the machine learning model, of the condition deterioration associated with the power transformer system further comprises: calculating a first rate of change associated with an event in the power transformer system, wherein the event occurs during a first time period; A rate of change of dissolved hydrogen associated with insulation degradation of a power transformer, based on data for a prediction period (Zyglowicz, ¶ 66). calculating, using the reconstructed machine learning model, a first risk associated with the power transformer system based at least in part on the first rate of change; A health profile is generated based on the prediction period, including a risk, e.g., a risk of a fire event if a maintenance action correcting the degraded insulation is not performed within a timeframe (Zyglowicz, ¶¶ 74–76). The health profile is generated using a machine learning algorithm generated based on historical data (Zyglowicz, ¶ 31). calculating a second rate of change associated with the power transformer system, wherein the second rate of change is determined over a second time period; The industrial asset [transformer] may be monitored for multiple assessment/prediction periods (Zyglowicz, ¶ 61). calculating, using the reconstructed machine learning model, a second risk associated with the power transformer system based at least in part on the second rate of change; and The multiple assessment/prediction periods may be used to generate multiple health profiles (Zyglowicz, ¶ 61). determining the condition deterioration associated with the power transformer system based at least in part on the first risk and the second risk. The health profile may indicate, e.g., degradation [deterioration] of the insulation of a transformer (Zyglowicz, ¶¶ 65–67).
Regarding dependent claim 13, the rejection of claim 12 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the first risk and the second risk are calculated via a dissolved gas analyzer analytics engine, and wherein the dissolved gas analyzer analytics engine utilizes the reconstructed machine learning model, historical data associated with the power transformer system, and real-time data associated with the power transformer system. The data used by the machine learning algorithm may include dissolved gas readings (Zyglowicz, ¶¶ 46, 52, 53, 57, 62–65). The data used may be past [historical] data, present [real-time] data, or estimated future data (Zyglowicz, ¶ 52).
Regarding dependent claim 14, the rejection of claim 8 is incorporated and Klein/Hunt/Zyglowicz further teaches: further comprising: outputting a health status indicative of the condition deterioration to an operator. A health profile [health status indicator] is generated, including a recommended maintenance plan for the user to follow (Zyglowicz, ¶ 61). The health profile includes an indication of, e.g., degraded insulation (Zyglowicz, ¶ 65).
Regarding independent claim 15, Klein teaches [a] power transformer system, comprising: […] determine a quantum of historical data associated with a machine learning model; The machine learning model may use historical data, e.g., historical malware signatures, to make predictions (Klein, ¶ 94). Machine learning capabilities of network devices are compared to a machine learning request, including memory capacity/availability, storage capacity, data transmission rate, etc. [capabilities related to amounts/quanta of data] (Klein, ¶¶ 49–53). The machine learning request is routed based on the processing load, including memory consumption (Klein, ¶ 135). determine an order associated with the machine learning model; The machine learning capabilities include the ability to process neural networks of a particular size or topology, models of a particular type, hyperparameters, etc. [complexity/order of the models] (Klein, ¶¶ 48, 50, 53, 159, 163). determine whether a latency associated with the machine learning model is critical; The machine learning request may have a QoS [quality of service] requirement, including a latency requirement (Klein, ¶ 135). select a server from a plurality of servers based at least in part on the quantum of historical data, the order, and the latency; and The machine learning request is routed to a network device based on a comparison of the request and the capabilities and characteristics of the available devices (Klein, ¶¶ 49, 159, 167, 168). […]
Klein teaches training the machine learning model using the server (Klein, ¶¶ 79, 80) but does not expressly teach reconstructing a model from a weight matrix. However, Hunt teaches: receive from the server, one or more structural details of the machine learning model and a corresponding weight matrix representation of the machine learning model; A plurality of neural network models [machine learning models] are stored as weight matrices and tensors (Hunt, ¶ 94). The storage media may be part of a distributed computing system and accessed using a communication interface (Hunt, ¶ 26). reconstruct the machine learning model from the one or more structural details of the machine learning model and the corresponding weight matrix representation; and A plurality of neural network models [machine learning models] are initialized by using weight matrices and tensors learned from a training step to reconstruct the models (Hunt, ¶ 94).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein with those of Hunt. Doing so would have been a matter of simple substitution of one known element [the model storage method of Klein] for another [the model storage method of Hunt] to obtain predictable results [a method of selecting a server based on a latency of a model, wherein the model is stored as a weight matrix].
Klein/Hunt teaches routing machine learning requests to servers based on aspects of the requests compared to the servers, but does not expressly teach performing machine learning related to power transformers. However, Zyglowicz teaches: a power transformer; and An industrial asset, such as a transformer of a power system (Zyglowicz, ¶¶ 24–25). a dissolved gas analyzer analytics engine, wherein the dissolved gas analyzer analytics engine is configured to: A model that analyzes dissolved gas concentrations (Zyglowicz, ¶¶ 52, 53, 57). determine, using the reconstructed machine learning model, a condition deterioration associated with the power transformer. The health profile may indicate, e.g., degradation [deterioration] of the insulation of a transformer (Zyglowicz, ¶¶ 65–67).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein/Hunt with those of Zyglowicz. Doing so would have been a matter of applying a known technique [routing ML requests to optimal servers] to a known method ready for improvement [the ML method for power transformers] to yield predictable results [a ML method for power transformers wherein the ML is routed to optimal servers based on comparing the models to the servers].
Regarding dependent claim 17, the rejection of claim 15 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the quantum of historical data associated with the machine learning model comprises one of a low quantum of historical data, a medium quantum of historical data, or a high quantum of historical data. The model may be a large-scale model with distributed input data [a high quantum of data] and processed by multiple network devices (Klein, ¶ 57). Machine learning capabilities of network devices are compared to a machine learning request, including memory capacity/availability, storage capacity, data transmission rate, etc. [capabilities related to amounts/quanta of data] (Klein, ¶¶ 49–53). [The requests are matched to servers based in part on their memory/storage capacity, therefore some requests require a lower capacity and some require a higher capacity.]
Regarding dependent claim 18, the rejection of claim 15 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the order associated with the machine learning model comprises one of a reduced order, a moderately high order, or a complex order. The machine learning processing capabilities include a size of neural network, number of layers or neurons, etc. [therefore machine learning requests are matched based on these aspects of model order/complexity, and the model orders may be higher or lower] (Klein, ¶ 48).
Regarding dependent claim 19, the rejection of claim 15 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the determination, using the reconstructed machine learning model, of the condition deterioration associated with the power transformer system comprises: calculating a first rate of change associated with an event in the power transformer system, wherein the event occurs during a first time period; A rate of change of dissolved hydrogen associated with insulation degradation of a power transformer, based on data for a prediction period (Zyglowicz, ¶ 66). calculating, using the reconstructed machine learning model, a first risk associated with the power transformer system based at least in part on the first rate of change; A health profile is generated based on the prediction period, including a risk, e.g., a risk of a fire event if a maintenance action correcting the degraded insulation is not performed within a timeframe (Zyglowicz, ¶¶ 74–76). The health profile is generated using a machine learning algorithm generated based on historical data (Zyglowicz, ¶ 31). calculating a second rate of change associated with the power transformer system, wherein the second rate of change is determined over a second time period; The industrial asset [transformer] may be monitored for multiple assessment/prediction periods (Zyglowicz, ¶ 61). calculating, using the reconstructed machine learning model, a second risk associated with the power transformer system based at least in part on the second rate of change; and The multiple assessment/prediction periods may be used to generate multiple health profiles (Zyglowicz, ¶ 61). determining the condition deterioration associated with the power transformer based at least in part on the first risk and the second risk. The health profile may indicate, e.g., degradation [deterioration] of the insulation of a transformer (Zyglowicz, ¶¶ 65–67).
Regarding dependent claim 20, the rejection of claim 15 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the dissolved gas analyzer analytics engine is further configured to: output a health status indicative of the condition deterioration to an operator. A health profile [health status indicator] is generated, including a recommended maintenance plan for the user to follow (Zyglowicz, ¶ 61). The health profile includes an indication of, e.g., degraded insulation (Zyglowicz, ¶ 65).
Claims 9 and 16 are rejected under 35 U.S.C. § 103 as being unpatentable over Klein et al. (US 2022/0239758 A1) [hereinafter Klein] in view of Hunt et al. (US 2018/0211153 A1) [hereinafter Hunt] and Zyglowicz et al. (US 2014/0365191 A1) [hereinafter Zyglowicz], further in view of Hayashi et al. (US 2023/0191474 A1) [hereinafter Hayashi].
Regarding dependent claim 9, the rejection of claim 8 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the server is one of a cloud server or […], wherein reconstructing the machine learning model is performed by a model serializer of an embedded edge server, and further wherein the reconstructed machine learning model is executed by a machine learning model training and execution module of the embedded edge server. The network devices/nodes may be servers, including cloud servers and/or edge servers (Klein, ¶¶ 38–40, 52). The network devices include an integrated machine learning core having a pre-processing module, which includes a training component [training module] and local inference module [execution module] (Klein, ¶¶ 77–81).
Klein/Hunt/Zyglowicz teaches performing machine learning functions on edge servers and cloud servers, but does not expressly teach FOG servers. However, Hayashi teaches: a Free and Open-Source Ghost (FOG) server A machine learning device predicts a quality of output of a connected device; the machine learning device may be a FOG server (Hayashi, ¶ 82).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein/Hunt/Zyglowicz with those of Hayashi. Doing so would have been a matter of applying a known technique [hosting a ML model on a FOG server] to a known method ready for improvement [the ML model hosting of Klein] to yield predictable results [a method of routing ML models to cloud, edge, and FOG servers].
Regarding dependent claim 16, the rejection of claim 15 is incorporated and Klein/Hunt/Zyglowicz further teaches: wherein the server is one of a cloud server or […], wherein reconstructing the machine learning model is performed by a model serializer of an embedded or edge server, and further wherein the reconstructed machine learning model is executed by a machine learning model training and execution module of the embedded edge server. The network devices/nodes may be servers, including cloud servers and/or edge servers (Klein, ¶¶ 38–40, 52). The network devices include an integrated machine learning core having a pre-processing module, which includes a training component [training module] and local inference module [execution module] (Klein, ¶¶ 77–81).
Klein/Hunt/Zyglowicz teaches performing machine learning functions on edge servers and cloud servers, but does not expressly teach FOG servers. However, Hayashi teaches: a Free and Open-Source Ghost (FOG) server A machine learning device predicts a quality of output of a connected device; the machine learning device may be a FOG server (Hayashi, ¶ 82).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine the teachings of Klein/Hunt/Zyglowicz with those of Hayashi. Doing so would have been a matter of applying a known technique [hosting a ML model on a FOG server] to a known method ready for improvement [the ML model hosting of Klein] to yield predictable results [a method of routing ML models to cloud, edge, and FOG servers].
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 C.F.R. § 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 C.F.R. § 1.17(a)) pursuant to 37 C.F.R. § 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.
The prior art made of record and not relied upon is considered pertinent to Applicant's disclosure.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Tyler Schallhorn whose telephone number is 571-270-3178. The examiner can normally be reached Monday through Friday, 8:30 a.m. to 6 p.m. (ET).
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, Tamara Kyle can be reached at 571-272-4241. 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 the USA or Canada) or 571-272-1000.
/Tyler Schallhorn/Examiner, Art Unit 2144
/TAMARA T KYLE/Supervisory Patent Examiner, Art Unit 2144
Prosecution Insights