DETAILED ACTION
This Action is responsive to Claims filed 01/09/2026.
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 .
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 01/09/2026 has been entered.
Status of the Claims
Claims 1-4, 6-9, 14-17, 19, and 29-30 have been amended. Claims 5, 13, 18, 20-24, 26-28, and 31-35 were previously cancelled. Claims 1-4, 6-12, 14-17, 19, 25, and 29-30 are currently pending.
Response to Arguments
Applicant's arguments, see Pages 11-13, filed 01/09/2026 regarding the 35 U.S.C. 101 Rejection of Claims 1-4, 6-12, 14-17, 19, 25, and 29-30 have been fully considered but they are not persuasive.
Regarding the Applicant’s comparison of the present Application to the recent Desjardins decision: The Examiner fails to see how the instant Application and Desjardins are similar. Desjardins pertains to catastrophic memory forgetting, and the fact pattern of the instant Application differs from this subject. The Examiner considers all relevant statutes equally when making a determination on patentability.
The Examiner respectfully disagrees with the Applicant’s characterization of the analysis and submits the 35 U.S.C. 101 Rejection should be maintained, under the claims’ present drafting. The Examiner does not disagree that an essential part of implementing the proposed improvement the Applicant alleges would be sending and receiving the specific data, and performing calculations or estimations on this data to ascertain an optimal outcome. However, as presently drafted, the specific improvement the Applicant alleges is not rooted in an additional element reciting significantly more than the abstract idea mental process steps. The improvement is not rooted in sending or receiving this specific or relevant data, as the sending and receiving of said data is recited highly generally. The improvement, the Examiner contends, then must be rooted in the manipulations, calculations, or estimations and subsequent determination made off of this data, in order to realize this improvement. There is no efficiency improvement realized until this determination is made, and subsequently executed.
As presently drafted, the “estimating…”, “comparing…”, and “determining…” steps are not claimed in such as a way as to preclude their performance by a human mind with or without the aid of pen and paper. In fact, the Applicant has amended the only structural implementation of these limitations to be more generic (a generic device versus a reference to a neural network, see MPEP 2106.05(f)(2) regarding generic computer components). As presently drafted, the steps not interpretable as abstract idea mental process steps are recited highly generally without specific implementation or structure intrinsically tying the abstract idea mental process steps to them in such a way that precludes a human mind from performing them with or without the aid of pen and paper. Therefore, the Examiner contends the proposed improvement is rooted in the implementation of an algorithmic set of abstract idea mental process steps, which, per MPEP 2106.05(a), the specific improvement cannot come from. See the updated 35 U.S.C. 101 Rejection below.
Applicant’s arguments, see Pages 13-15, filed 01/09/2026, with respect to the rejection(s) of claim(s) 1-4, 6-12, 14-17, 19, 25, and 29-30 under 35 U.S.C. 102(a)(1) have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of 35 U.S.C. 103.
Claim Objections
Claims 1, 14, and 29 objected to because of the following informalities:
The newly amended portion of the “estimating…” step recites “distributed neural already…” instead of “neural network”
Appropriate correction is required.
Claim Rejections - 35 USC § 101
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1-4, 6-12, 14-17, 19, 25, and 29-30 rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more; and because the claims as a whole, considering all claim elements both individually and in combination, do not amount to significantly more than the abstract idea, see Alice Corporation Pty. Ltd. v. CLS Bank International, et al, 573 U.S. (2014). In determining whether the claims are subject matter eligible, the Examiner applies the 2019 USPTO Patent Eligibility Guidelines. (2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50, Jan. 7, 2019.)
Step 1:
Claims 1-4 and 6-12 recite a method for dynamic load distribution for a distributed neural network which falls under the statutory category of a process. Claims 14-17, 19, 25 recite an apparatus for dynamic load distribution for a distributed neural network, which falls under the statutory category of a machine. Claims 29 and 30 recite a system for dynamic load distribution for a distributed neural network, which falls under the statutory category of a machine.
Step 2A – Prong 1:
Claim 1 recites an abstract idea, law of nature, or natural phenomenon. The limitations of “estimating, from said sensing, channel properties of the wireless communication channel, wherein the estimated channel properties comprise an effective channel bitrate;”, “estimating, by the device and based on at least the estimated channel properties, an energy usage for transmitting layer output of at least one processed layer of the distributed neural already processed by the device to a cloud service of the neural network for processing of all of the non-processed layers;”, “comparing, by the device, the estimated energy usage for processing all of the non-processed layers in the device with the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service;”, and “determining whether to process the non-processed layers in the device according to a rule by which the device determines to process all of the non-processed layers in the device when the estimated…such that control of energy expended…” under the broadest reasonable interpretation, cover a mental process including an observation, evaluation, judgment or opinion that could be performed in the human mind or with the aid of pencil and paper. These limitations therefore fall within the mental process group.
Step 2A – Prong 2:
The additional elements of claim 1 do not integrate the abstract idea into a judicial exception. The claim recites the additional elements “a device”, “an energy usage”, “a wireless communication channel”, “channel properties”, “an effective channel bitrate”, “a cloud service”, “estimated energy usage”, and “a rule” are recognized as generic computer components recited at a high level of generality. Although it has and executes instructions to perform the abstract idea itself, this also does not serve to integrate the abstract idea into a practical application as it merely amounts to instructions to "apply it." (See MPEP 2106.04(f)(2) indicating mere instructions to apply an abstract idea does not amount to integrating the abstract idea into a practical application).
The additional elements recited in the limitations “a distributed neural network”, “a plurality of layers”, “the neural network”, “one non-processed layer”, “layer output”, and “one processed layer” are recognized as non-generic computer components, however, they are found to generally link the abstract idea to a particular technological field.
The additional elements recited in the limitations “obtaining, by the device, an energy usage for processing all non-processed layers of the distributed neural network in the device, the non-processed layers compromising two or more layers;”, “sensing a wireless communication channel…”, and “transmitting the layer output of the at least one processed layer to the cloud service is equal or greater than the estimated energy usage for processing all of the non-processed layers in the device and instead determines to transmit the layer output of the at least one processed layer to the cloud service for processing of all of the non-processed layers when the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service is less than the estimated energy usage for processing all of the non-processed layers in the device,” Amount to mere insignificant extra-solution activity (See MPEP 2106.05(g) for sending and receiving data).
Step 2B:
The only limitation on the performance of the described method is a limitation reciting “a device”, “an energy usage”, “an energy usage”, “a wireless communication channel”, “channel properties”, “an effective channel bitrate”, “a cloud service”, “estimated energy usage”, and “a rule” These elements are insufficient to transform a judicial exception to a patentable invention because the recited elements are considered insignificant extra-solution activity (generic computer system, processing resources, links the judicial exception to a particular, respective, technological environment). The claim thus recites computing components only at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components; mere instructions to apply an exception using a generic computer component cannot provide an inventive concept (see MPEP 2106.05(f)).
The additional elements of “a distributed neural network”, “a plurality of layers”, “the neural network”, “one non-processed layer”, “layer output”, and “one processed layer” are recognized as non-generic computer components, however, they are found to generally link the abstract idea to a particular technological field.
The additional elements recited in the limitations “obtaining, by the device, an energy usage for processing all non-processed layers of the distributed neural network in the device, the non-processed layers compromising two or more layers;”, and “transmitting the layer output of the at least one processed layer to the cloud service is equal or greater than the estimated energy usage for processing all of the non-processed layers in the device and instead determines to transmit the layer output of the at least one processed layer to the cloud service for processing of all of the non-processed layers when the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service is less than the estimated energy usage for processing all of the non-processed layers in the device,” Are found to be well understood, routine or conventional activities (See MPEP 2106.05(d)(II)(i)).
Taken alone or in ordered combination, these additional elements do not amount to significantly more than the above-identified abstract idea. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide conventional computer implementation.
For the reasons above, claim 1 is rejected as being directed to non-patentable subject matter under §101. This rejection applies equally to independent claims 14 and 29.
Claim 14 recites similar limitations to claim 1, with the exception of the recitation of “an apparatus”, “a transceiver”, “a memory”, “executable instructions”, and “one or more processors” generic computer components. The limitations of claim 8 have been evaluated under step 2A Prong 2 and reevaluated under step 2B and found to be recited at high levels of generality.
Claim 29 recites similar limitations to claim 1, with the exception of the recitation of “a system” generic computer components. The limitations of claim 8 have been evaluated under step 2A Prong 2 and reevaluated under step 2B and found to be recited at high levels of generality. Claim 29 further recites additional elements “circuitry” These additional elements have been evaluated under step 2A Prong 2 and reevaluated under step 2B and found to generally link the abstract idea to a particular technological field.
Dependent Claims:
Claim 2 (claims 15 and 30) recites the aforementioned additional elements of claim 1 as well as a mental process step “determining, by the device, at least one layer output of the at least one processed layer for processing the non-processed layers.”
Claim 3 (claim 16) recites the aforementioned additional elements of claim 1 as well as a mental process step “determining multiple layer outputs of multiple processed layers for processing the non-processed layers.”
Claim 4 (claim 17) recites refinements to the aforementioned mental process steps. As well as the limitation “receiving an input, in the device, wherein the input comprises any one of image data, voice data, video data, and temperature data.” This limitation has been evaluated under Step 2A – Prong 2 and reevaluated under Step 2B and has been found to be a mere data transmittal step.
Claim 6 (claim 19) recites data manipulation in the form of “encoding and/or compressing the layer output of the at least one processed layer when it is determined to transmit the layer output of the at least one processed layer to the cloud service of the neural network for processing the non-processed layers.” This limitation has been evaluated under Step 2A – Prong 2 and reevaluated under Step 2B and has been found to be mere instructions to apply the judicial exception (See MPEP 2106.05(d)).
Claim 7 recites a mental process step in the form of “recording the estimated energy usage for processing the non-processed layers layer-wise in the device in response to estimating, by the device, the energy usage for processing the non-processed layers in the device.” This limitation has been evaluated under Step 2A – Prong 2 and reevaluated under Step 2B and has been found to be mere insignificant extra-solution activity (See WURC examples MPEP 2106.05(d)(II)(iv)).
Claim 8 recites additional elements “multiply-accumulate operations”, “memory accesses”, “non-linear activation functions”, “normalization”, “padding”, and “pooling” These additional elements have been evaluated under Step 2A – Prong 2 and reevaluated under Step 2B and has been found to generally link the abstract idea to a particular technological field.
Claim 9 recites a mere data manipulation step “…processing of the non-processed layers comprises inference processing.” as part of the data manipulated in the abstract idea of claim 1.
Claim 10 recites a refinement of the additional elements of claim 1. The claimed “resource constrained device” additional element has been evaluated under Step 2A – Prong 2 and reevaluated under Step 2B and found to be recited at a high level of generality.
Claim 11 recites a refinement of the additional elements of claim 1. The claimed “sensor” additional element has been evaluated under Step 2A – Prong 2 and reevaluated under Step 2B and found to be recited at a high level of generality.
Claim 12 (claim 25) recites a refinement of the additional elements of claim 1. The claimed “edge cloud service” additional element has been evaluated under Step 2A – Prong 2 and reevaluated under Step 2B and found to be recited at a high level of generality.
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.
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 CFR 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.
Claim(s) 1-4, 6-12, 14-17, 19, 25, and 29-30 is/are rejected under 35 U.S.C. 103 as being unpatentable over Eshratifar et al. (JointDNN: An Efficient Training and Inference Engine for Intelligent Mobile Cloud Computing Services, 2019), hereinafter Eshratifar and Kang et al. (Neurosurgeon: Collaborative Intelligence Between the Cloud and Mobile Edge, 2017), hereinafter Kang.
In regards to claim 1: The present invention claims: “A method of dynamically controlling energy expended in a device when operating a distributed neural network that is distributed between the device and a cloud service, wherein processing by the distributed neural network comprises processing a plurality of layers such that one or more of the plurality of layers are processed in the device and a remaining number of the plurality of layers are processed by the cloud service, the method comprising the steps of:” Eshratifar teaches “we propose an efficient, adaptive, and practical engine, JointDNN, for collaborative computation between a mobile device and cloud for DNNs in both inference and training phase.” and “Given the DNN architecture, we investigate the efficiency of processing some layers on the mobile device and some layers on the cloud server. We provide optimization formulations at layer granularity for forward and backward propagation in DNNs…” (Page 1, Abstract). Eshratifar also teaches implementations that bounce back and forth between the mobile device, cloud, and back, as well as implementations that start exclusively on mobile, and end exclusively on the cloud. (Page 573, Figure 11, mapping to “a remaining number of the plurality of layers are processed by the cloud service”).
“obtaining, by the device, an energy usage for processing all non-processed layers of the distributed neural network in the device, the non-processed layers comprising two or more layers;” Eshratifar teaches a graph model with 4 possible cases for what an edge between nodes represents, including “1) A transition from the mobile to the mobile, which only includes the mobile computation cost (MEi,j)” (Page 4, Starting at “In this graph…”). See the previous two paragraphs where “…costs can refer to either latency or energy.”
“sensing a wireless communication channel and estimating, from said sensing, channel properties of the wireless communication channel, wherein the estimated channel properties comprise an effective channel bitrate;” Eshratifar teaches, in the context of their shortest path algorithm, “However, the problem of the shortest path subjected to constraints is NP-Complete [22]. For instance, assuming our standard graph is constructed for energy and we need to find the shortest path subject to the constraint of the total latency of that path is less than a time deadline (QoS). However, there is an approximation solution to this problem, “LARAC” algorithm [23], the nature of our application does not require to solve this optimization problem frequently, therefore, we aim to obtain the optimal solution. We can constitute a small look-up table of optimization results for a different set of parameters (e.g., network bandwidth, cloud server load, etc.). We provide the ILP formulations of DNN partitioning in the following sections.” (Page 568). The aforementioned network parameters are later seen on page 570 being used in Algorithm 1, which defines Eshratifar’s scheduling. Eshratifar also teaches solving their scheduling algorithm for multiple uplink and download speeds (mapping to bitrate) (Page 570, Column 1).
“estimating, by the device and based on at least the estimated channel properties, an energy usage for transmitting layer output of at least one processed layer of the distributed neural already processed by the device to a cloud service of the neural network for processing of all of the non-processed layers;” See how the aforementioned algorithm 1 incorporates channel parameters. Another case for an edge in Eshratifar’s graph includes “3) A transition from the mobile to the cloud, which includes the mobile computation cost and uploading cost of the inputs of the next node (EUi, j = MEi, j + UIDj+1)” (Page 4, Starting at “In this graph…”).
While Eshratifar teaches scenarios where the optimal solution to the shortest path calculation may be processing one or more layers on an edge device, and all remaining layers on the cloud (as depicted in Figure 11, for example), Eshratifar fails to explicitly teach requiring that the cloud service complete the processing, rather than sending back to the mobile device, which the claims dictate. Although the Examiner notes Eshratifar does teach “State-of-the-art work for collaborative computation of DNNs [14] only considers one offloading point, assigning computation of its previous layers and next layers on the mobile and cloud platforms, respectively. We show that this approach is non-generic and fails to be optimal, and introduced a new method granting the possibility of computation on either platform for each layer independent of other layers.” (Page 566, right column). Eshratifar also teaches, primarily on Page 567, various means by which the energy costs may be calculated, including as a whole, by groups, or layer-by-layer.
The reference cited by Eshratifar [14], Kang, in a similar field of endeavor of mobile-cloud collaborative computation, however, is designed with a single partition point in mind (See Kang Figure 1c and Section 4.3 at least) where computation of the neural network does not leave the cloud once the optimal partition point is found.
A person of ordinary skill in the art, at least at the time of Eshratifar’s writing, would have been aware of the state of the art at the time before the Applicant’s effective filing date. A person of ordinary skill in the art would have been aware of the optimality (or in Eshratifar’s case, the lack thereof) of finding a single partition point, and would have reasonably used methods similar to Kang and/or Eshratifar to find an optimal point, based on energy consumption computed layer-by-layer, by groups, or by the whole, to partition the neural network before processing the non-processed layers on the cloud service.
The mapping of the following limitations is made with the knowledge that a person of ordinary skill in the art may have used known methods from Kang and/or Eshratifar in order to arrive at this partition point and process “all non-processed layers” on the cloud service.
“comparing, by the device, the estimated energy usage for processing all of the non-processed layers in the device with the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service;” Eshratifar teaches, in reference to the JointDNN graph model, “Under this formulation, we can transform the computation scheduling problem to finding the shortest path from S to F” (Page 4, Starting at “In this graph…”, mapping finding a shortest path, based on edge cost, when the edges represent either of the cases above, to the comparison step). See also Table 1 on page 4, where the description of various variables is provided, including the cost (can be energy per above citation) of executing a layer on either the mobile device (ME) or cloud (CE), as well as ED and EU which represent transitions from cloud to mobile and back. Finding an optimal path through the JointDNN graph would require comparison between these values.
“determining whether to process the non-processed layers in the device according to a rule by which the device determines to process all of the non-processed layers in the device when the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service is equal or greater than the estimated energy usage for processing all of the non-processed layers in the device and instead determines to transmit the layer output of the at least one processed layer to the cloud service for processing of all of the non-processed layers when the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service is less than the estimated energy usage for processing all of the non-processed layers in the device,” See above how Eshratifar Page 4 and Table 1 (ME and EU) teaches JointDNN optimizing the shortest path to reduce the cost of computing layers on an edge device or cloud service. If the cost of example 3) of an edge (Page 4) is greater than example 1) of an edge (Page 4), then the optimal solution would include processing the layer on the edge device. Eshratifar Page 4 and Table 1 (ME and EU) teaches JointDNN optimizing the shortest path to reduce the cost of computing layers on an edge device or cloud service. If the cost of example 1) of an edge (Page 4), is greater than example 3) of an edge (Page 4), then the optimal solution would include transmitting the layer output to the cloud service for processing. Eshratifar also teaches “During online training, the huge communication overhead of transmitting the updated weights will be added to the total cost. Therefore, to avoid downloading this large data, only a few back-propagation steps are computed in the cloud server.” (Page 571) and “They show how the computations in DNN is divided between the mobile and the cloud. As can be seen, discriminative models (e.g., AlexNet), inference follows a mobile-cloud pattern and training follows a mobile-cloud-mobile pattern. The intuition is that the last layers are computationally intensive (fully connected layers) but with small data sizes, which require a low communication cost, therefore, the last layers tend to be computed on the cloud.” (Page 572).
“such that control of energy expended by the distributed neural network is dynamically based, at least in part, on the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service of the neural network for processing.” See above where Eshratifar teaches using channel properties or estimated energies to find a shortest, most efficient path in executing a DNN on a mobile device and a cloud computing service.
In regards to claim 2: The present invention claims: “determining, by the device, at least one layer output of the at least one processed layer for processing the non-processed layers.” See above where Kang teaches finding a partition point in order to finish processing the remaining layers on the cloud service. Eshratifar teaches “a sequence of distinct layers with a linear topology as depicted in Figure 4. Layers are executed sequentially, with output data generated by one layer feeds into the input of the next one.” (Page 4, Starting at “First we assume…”, Figure 4).
In regards to claim 3: The present invention claims: “determining multiple layer outputs of multiple processed layers for processing the non-processed layers” See above where Kang teaches finding a partition point in order to finish processing the remaining layers on the cloud service. Eshratifar teaches “a sequence of distinct layers with a linear topology as depicted in Figure 4. Layers are executed sequentially, with output data generated by one layer feeds into the input of the next one.” (Page 4, Starting at “First we assume…”, Figure 4).
In regards to claim 4: The present invention claims: “receiving an input, in the device, wherein the input comprises any one of image data, voice data, video data, and temperature data.” See above where Kang teaches finding a partition point in order to finish processing the remaining layers on the cloud service. Eshratifar teaches the use of their system in speech recognition and image processing benchmarks (Page 7, Starting at Discriminative Neural Networks).
In regards to claim 6: The present invention claims: “encoding and/or compressing the layer output of the at least one processed layer when it is determined to transmit the layer output of the at least one processed layer to the cloud service of the neural network for processing the non-processed layers.” See above where Kang teaches finding a partition point in order to finish processing the remaining layers on the cloud service. Eshratifar teaches compressing the output of layers to conserve energy and delay in Section 4.2 Layer Compression (Pages 10-11).
In regards to claim 7: The present invention claims: “recording the estimated energy usage for processing the non-processed layers layer-wise in the device in response to estimating, by the device, the energy usage for processing the non-processed layers in the device.” See above where Kang teaches finding a partition point in order to finish processing the remaining layers on the cloud service. Eshratifar teaches “Our standard graph model has a memory of one which is the very previous layer. We provide a method to transform the computation graph of this type of network to our standard model, JointDNN graph. In this regard, we add two additional chains of size k − 1, where k is the number of nodes in the residual block (3 in Figure 7). One chain represents the case of computing layer 2 on the mobile and the other one represents the case of computing layer 2 on the cloud. In Figure 7, we have only shown the weights that need to be modified, where D2 and U2 are the cost of downloading and uploading the output of layer 2, respectively. By solving the shortest path problem in JointDNN graph model, we can obtain the optimal scheduling of inference in DNNs.” (Page 4, mapping the solving of shortest path problem to the recording of relevant information necessary to solve said shortest path problem).
In regards to claim 8: The present invention claims: “…the energy usage for processing the non-processed layers in the device of the neural network comprises energy used for any one of multiply-accumulate operations, memory accesses, non-linear activation functions, normalization, padding, and pooling.” See above where Kang teaches finding a partition point in order to finish processing the remaining layers on the cloud service. Eshratifar teaches tracking the energy usage of a DNN, and the building blocks of DNNs including pooling, activation, and normalization layers (Page 3).
In regards to claim 9: The present invention claims: “…the processing of the non-processed layers comprises inference processing.” See above where Kang teaches finding a partition point in order to finish processing the remaining layers on the cloud service. Eshratifar teaches their system is for both inference and training (Abstract).
In regards to claim 10: The present invention claims: “…the device of the neural network is a resource constrained device.” Eshratifar teaches the edge devices can be mobile devices, which have limited storage and power capabilities (Abstract).
In regards to claim 11: The present invention claims: “…wherein the resource constrained device comprises a sensor.” Eshratifar teaches “We used Jetson TX2 module developed by NVIDIA® [3], a fair representative of mobile computation power as our mobile device. This module enables efficient implementation of DNN applications used in products such as robots, drones, and smart cameras. It is equipped with NVIDIA Pascal®GPU with 256 CUDA cores and a shared 8 GB 128 bit LPDDR4 memory between GPU and CPU. To measure the power consumption of the mobile platform, we used INA226 power sensor [17].” (Page 8, Section 3.2, mapping to the inclusion of mobile devices, robots, drones, and smart camera, all of which possess a plethora of sensors).
In regards to claim 12: The present invention claims: “wherein the cloud service of the neural network comprises an edge cloud service.” Eshratifar teaches the devices engaged with the cloud service can be mobile devices (Abstract). Eshratifar also teaches “Most recently, scalable distributed hierarchy structures between end-user device, edge, and cloud have been suggested [39] which are specialized for DNN applications.” (Page 1, Starting at “Automatic partitioning…”).
In regards to claims 14-17, 19, and 25: Claims 14-17, 19, and 25 recite similar limitations to Claims 1-4, 6, and 12, save for “An apparatus for dynamically controlling energy expended in a device when operating a distributed neural network that is distributed between the device and a cloud service, wherein processing by the distributed neural network comprises processing a plurality of layers such that one or more of the plurality of layers are processed in the device and a remaining number of the plurality of layers are processed by the cloud service, the apparatus comprising:” therefore, Claims 14-17, 19, and 25 are similarly rejected in view of Eshratifar and Kang, both because the claimed apparatus relies on the above rejected method, and because Eshratifar also teaches their method being tested on requisite hardware (Page 2, Starting at “To present realistic results…”).
In regards to claims 29 and 30: Claims 29 and 30 recite similar limitations to Claims 1 and 2, save for “A system for dynamically controlling energy expended in a device when operating a distributed neural network that is distributed between the device and a cloud service, wherein processing by the distributed neural network comprises processing a plurality of layers such that one or more of the plurality of layers are processed in the device and a remaining number of the plurality of layers are processed by the cloud service, the system comprising:” and the recitation of circuitry to perform the method steps, therefore, Claims 29 and 30 are similarly rejected in view of Eshratifar and Kang, both because the claimed system relies on the above rejected method, and because Eshratifar also teaches their method being tested on requisite hardware (Page 2, Starting at “To present realistic results…”).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GRIFFIN T BEAN whose telephone number is (703)756-1473. The examiner can normally be reached M - F 7:30 - 4:30.
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, Li Zhen can be reached at (571) 272-3768. 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.
/GRIFFIN TANNER BEAN/ Examiner, Art Unit 2121
/Li B. Zhen/ Supervisory Patent Examiner, Art Unit 2121