DETAILED ACTION
This Office action is in response to communications filed on August 5, 2025 and August 29, 2025 for Application No. 17/566,624, in which claims 1-20 are presented for examination. The Amendments filed on August 5, 2025 and August 29, 2025 have been entered, where, collectively, claims 1, 3-4, 7-8, 10-11, 14, 16-17, and 20 are amended.
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 .
Information Disclosure Statement
The information disclosure statement submitted on 05/23/2025 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure was considered by the examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to abstract ideas without significantly more.
Regarding Claim 1:
Step 1: Claim 1 is a method claim. Therefore, Claims 1-7 are directed to a statutory category of eligible subject matter.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. Here, claim limitations are mental processes. Specifically, the claim recites
“learning, using a first machine learning system, an input transformation function that transforms input data for a second machine learning system, the learning using a minimization of a summation of a task loss and a post-activation density loss” (mental process – other than reciting “a first machine learning system”, amounts to using judgement and evaluate to develop a process to transform observed data under certain constraints, which may be aided by pen and paper) and
“transforming the input data using the learned input transformation function thereby reconfiguring the post-activation density” (mental process – amounts to using judgement and evaluation to transform observed data in a manner determined to reduce energy consumed for a task, which may be aided by pen and paper).
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional elements:
“a first machine learning system” (system recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components) and
“and reducing an amount of energy consumed for an inferencing task . . . carrying out the inferencing task on the transformed input data using the second machine learning system” (amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use).
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional elements:
“a first machine learning system” (mere instructions to apply the exception using generic computer components cannot provide an inventive concept);
“and reducing an amount of energy consumed for an inferencing task . . . carrying out the inferencing task on the transformed input data using the second machine learning system” (merely generally linking the exception to generic computer components cannot provide an inventive concept).
For the reasons above, Claim 1 is rejected as being directed to an abstract idea without significantly more. This rejection applies equally to dependent claims 2-7. The additional limitations of the dependent claims are addressed below.
Regarding Claim 2:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 2 depends on.
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional element:
“wherein the second machine learning system is implemented with a neural network” (use of neural network is recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components).
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional element:
“wherein the second machine learning system is implemented with a neural network” (mere instructions to apply the exception using generic computer components cannot provide an inventive concept).
Accordingly, Claim 2 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 3:
Step 2A Prong 1: See the rejection of Claim 2 above, which Claim 3 depends on. Here, the claim recites additional limitations that are abstract ideas:
“wherein the learning of the input transformation function comprises balancing the task loss and the post-activation density of the neural network at inference time based on a specified tradeoff factor” (mental process - amounts to using judgement and evaluate to develop a process to transform observed data to minimize specific losses for a known neural network given a specified tradeoff factor, which may be aided by pen and paper).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 3 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 4:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 4 depends on. Here, the claim recites additional limitations that are abstract ideas:
“repeating the learning of the input transformation function at different specified tradeoff factors to generate additional input transformation functions, wherein the transforming the input data using the learned input transformation is performed using a selected one of the input transformation functions corresponding to a selected tradeoff factor of the specified tradeoff factors” (mental process - amounts to using judgement and evaluate to develop multiple processes to transform observed data, network given a specified tradeoff factor, which may be aided by pen and paper).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 4 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 5:
Step 2A Prong 1: See the rejection of Claim 4 above, which Claim 5 depends on. Here, the claim recites additional limitations that are abstract ideas:
“obtaining the selection of one of the input transformation functions” (mental process - amounts to observing the functions and exercising judgement to form an opinion about which function to use).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 5 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 6:
Step 2A Prong 1: See the rejection of Claim 5 above, which Claim 6 depends on. Here, the claim recites additional limitations that are abstract ideas:
“the obtained selection is one of: the selected tradeoff factor, an energy level, an energy saving level, an identification of one of the input transformation functions, and an accuracy level” (mental process - amounts to observing the functions and exercising judgement to form an opinion about which function to use based on one of several corresponding factors).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 6 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 7:
Step 2A Prong 1: See the rejection of Claim 4 above, which Claim 7 depends on. Here, the claim recites additional limitations that are abstract ideas:
“in the learning step, minimizing the task loss ensures satisfactory accuracy for the inferencing task while minimizing the post-activation density loss ensures that a density of post-activations will be sufficiently small to attain the reduced energy consumption, and wherein each specified tradeoff factor serves to balance the task loss and the post-activation density loss to enable a tradeoff between the accuracy and the energy consumption” (mental process - amounts to using judgement and evaluate to develop multiple processes to transform observed data, while ensuring specified metrics are within a satisfactory range, which may be aided by pen and paper).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 7 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 8:
Step 1: Claim 8 is a product claim. Therefore, Claims 8-13 are directed to a statutory category of eligible subject matter.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. Here, the claim recites limitations that are substantially the same as the limitations of Claim 1. As a result, and as elaborated above, these limitations are abstract ideas because they are mental processes.
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional elements:
“[a] non-transitory computer readable medium comprising computer executable instructions which when executed by a computer cause the computer to perform the method” (medium recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components);
“a first machine learning system” (system recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components); and
“reducing an amount of energy consumed for an inferencing task . . . carrying out the inferencing task on the transformed input data using the second machine learning system” (amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use).
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional elements:
“[a] non-transitory computer readable medium comprising computer executable instructions which when executed by a computer cause the computer to perform the method” (mere instructions to apply the exception using generic computer components cannot provide an inventive concept);
“a first machine learning system” (mere instructions to apply the exception using generic computer components cannot provide an inventive concept); and
“reducing an amount of energy consumed for an inferencing task . . . carrying out the inferencing task on the transformed input data using the second machine learning system” (merely generally linking the exception to generic computer components cannot provide an inventive concept).
For the reasons above, Claim 8 is rejected as being directed to an abstract idea without significantly more. This rejection applies equally to dependent claims 9-13. The additional limitations of the dependent claims are addressed below.
Regarding Claim 9, the claim recites substantially the same limitations as Claim 2, in the form of a non-transitory computer readable medium. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 9 is rejected under the same rationale.
Regarding Claim 10, the claim recites substantially the same limitations as Claim 3, in the form of a non-transitory computer readable medium. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 10 is rejected under the same rationale.
Regarding Claim 11, the claim recites substantially the same limitations as Claim 4, in the form of a non-transitory computer readable medium. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 11 is rejected under the same rationale.
Regarding Claim 12, the claim recites substantially the same limitations as Claim 5, in the form of a non-transitory computer readable medium. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 12 is rejected under the same rationale.
Regarding Claim 13, the claim recites substantially the same limitations as Claim 6, in the form of a non-transitory computer readable medium. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 13 is rejected under the same rationale.
Regarding Claim 14:
Step 1: Claim 14 is an apparatus claim. Therefore, Claims 14-20 are directed to a statutory category of eligible subject matter.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. Here, the claim recites limitations that are substantially the same as the limitations of Claim 1. As a result, and as elaborated above, these limitations are abstract ideas because they are mental processes.
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional elements:
“[a]n apparatus comprising: a memory; and at least one processor, coupled to said memory, and operative to perform operations” (apparatus recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components);
“a first machine learning system” (system recited at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components); and
“reducing to reduce an amount of energy consumed for an inferencing task . . . carrying out the inferencing task on the transformed input data using the second machine learning system” (amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use).
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional elements:
“[a]n apparatus comprising: a memory; and at least one processor, coupled to said memory, and operative to perform operations” (mere instructions to apply the exception using generic computer components cannot provide an inventive concept);
“a first machine learning system” (mere instructions to apply the exception using generic computer components cannot provide an inventive concept); and
“reducing to reduce an amount of energy consumed for an inferencing task . . . carrying out the inferencing task on the transformed input data using the second machine learning system” (merely generally linking the exception to generic computer components cannot provide an inventive concept).
For the reasons above, Claim 14 is rejected as being directed to an abstract idea without significantly more. This rejection applies equally to dependent claims 15-20. The additional limitations of the dependent claims are addressed below.
Regarding Claim 15, the claim recites substantially the same limitations as Claim 2, in the form of an apparatus. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 15 is rejected under the same rationale.
Regarding Claim 16, the claim recites substantially the same limitations as Claim 3, in the form of an apparatus. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 16 is rejected under the same rationale.
Regarding Claim 17, the claim recites substantially the same limitations as Claim 4, in the form of an apparatus. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 17 is rejected under the same rationale.
Regarding Claim 18, the claim recites substantially the same limitations as Claim 5, in the form of an apparatus. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 18 is rejected under the same rationale.
Regarding Claim 19, the claim recites substantially the same limitations as Claim 6, in the form of an apparatus. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 19 is rejected under the same rationale.
Regarding Claim 20, the claim recites substantially the same limitations as Claim 7, in the form of an apparatus. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 20 is rejected under the same rationale.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 8-9, and 14-15 are rejected under 35 U.S.C. 103 as being unpatentable over Lai et al. (hereinafter Lai) (Patent Pub. No. US 2021/0011288 A1) in view of Bourdev et al. (hereinafter Bourdev) (Patent Pub. No. US 2018/0174047 A1), and Shumailov et al. (hereinafter Shumailov) (“Sponge Examples: Energy-Latency Attacks on Neural Networks”).
Regarding Claim 1, Lai teaches a method (Para. [0039], “systems and methods . . . for implementing multiple neural networks across multiple devices”) comprising:
learning, using a first machine learning system (Fig. 1A, which depicts a machine learning system; Para. [0045], “The input data 110 can include any type or form of data for configuring, tuning, training and/or activating a neural network 114”)
an input transformation function (Fig. 3A, where “First Device 302a” includes “Neural Network 114a”, which is within the broadest reasonable interpretation of an input transformation function)
that transforms input data for a second machine learning system (Para. [0040], “The first device can generate a reduced data set based on the input data and provide the reduced data set to the second device as an input to the neural network on the second device”, where “reduced” input data is within the broadest reasonable interpretation of transformed input data),
the learning being using a minimization of . . . [accuracy loss] and . . . [energy consumed] (Fig. 1A; Para. [0045], “training . . . the neural network 114 to allow the neural network to improve accuracy”; Para. [0039], “methods can be implemented to optimize neural network computations for energy savings”; Fig. 1A; Para. [0045], “Tuning or configuring can refer to or include training . . . [which can include] establishing the neural network . . . to be successful for the type of problem or objective desired”);
transforming the input data using the learned input transformation function (Para. [0141], “The process 350 includes generating a reduced data set using . . . a first neural network and the input data”; Para [0126], “the reduced data set 310 is smaller in size, compressed, reduced in size, etc., compared to the input data 110”)
thereby reconfiguring [the function for] . . . reducing an amount of energy consumed for . . . [a] task (FIG. 3B; Para. [0139], “a process 350 . . . for cooperatively implementing multiple neural networks across multiple devices . . . Advantageously, the process 350 reduces transmission between devices and enables the devices to reduce processing requirements, thereby facilitating energy savings); and
carrying out the . . . task on the transformed input data using the second machine learning system (Fig. 3b, Reference 364, “Identify one or more features of the reduced data set using (1) a second neural network and the reduced data set”, where identifying features is within the broadest reasonable interpretation of “carrying out the task”).
Lai does not teach . . . a summation of task loss . . . a post-activation density loss . . . the post-activation density and . . . inferencing . . . inferencing . . . (where Lai does not teach the learning should be based on minimizing a summation of a task loss and a post-activation density loss. Additionally, Lai does not teach the transforming of the input data should be specifically to alter the post-activation density for an inferencing task, which is the task carried out by the second machine learning system).
However, Bourdev teaches . . . [learning using a minimization of] a summation of task loss and . . . [an efficiency metric] (Para. [0037], “A first loss, hereafter referred to as a task loss, refers to the accuracy of the predicted ML task . . . A second loss . . . refers to an encoding efficiency . . . Therefore, the neural network model 145 is trained to tradeoff between the two losses”, where the two “losses” are combined and analyzed as a “tradeoff”, which, in light of the specification, is within the broadest reasonable interpretation of summation)
[, where the minimizing of the task loss is to ensure satisfactory accuracy] (Fig. 2; Para. [0009], “The ML task system calculates a task loss that represents the accuracy of the output of the task model”; Para. [0051], “The loss feedback module 170 backpropagates the task loss 265 to train the task model 150 such that the task model 150 can better predict a ML task output 275 that results in a smaller task loss”).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the use of machine learning based on minimizing accuracy loss and energy loss to develop an input transformation function of Lai, with learning based on minimizing a summation of a task loss and an efficiency metric of Bourdev, in order to base training off values, which represent future use outcomes, but can be generated, analyzed, and adjusted during training (Bourdev, Fig. 2; Para. [0037], “The loss feedback module 170 calculates the task loss 265 by comparing the predicted ML task output 275 and a ML task output label 255 of the training example”; Para. [0064], “The ML task system 110 determines 515 a codelength loss for each compressed representation”, where the system generates values during training to balance accuracy and efficiency of future tasks).
Additionally, Shumailov teaches [minimizing energy consumption by minimizing] a post-activation density loss (Pg. 215, Para. 3, “inputs that lead to less sparse activations will increase the number of operations and the number of memory accesses, and thus energy consumption”, where “sparse activations” is within the broadest reasonable interpretation of activation density; and if increased activation density is observed, and therefore energy consumed, switching to inputs that lead to less sparse activations, then decreased activation density, and therefore decreased energy consumed, is observed switching to inputs that lead to more sparse activations) . . .
[transforming input data to alter] the post-activation density and (Abstract, “adversaries can exploit carefully-crafted sponge examples, which are inputs designed to maximize energy consumption”; Figure 1; Pg. 216, Para. 3, “We then iteratively evolve the population pool [of sponge inputs] as is depicted in Figure 1”) . . .
[for a task of] inferencing . . . inferencing . . . (Abstract, “high energy costs of neural network training and inference”).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the use of machine learning based on minimizing a summation of a task loss and an energy loss efficiency metric to develop an input transformation function to transform input data for another machine learning model of Lai as previously modified by Bourdev with machine learning based on minimizing post-activation density loss of altered input data and use of the altered input data by a second model to carry out inferencing of Shumailov so the energy consumed for inferencing by the second machine learning system would be decreased (Shumailov, Pg. 215, Para. 3, A large number of ASIC neural network accelerators consequently exploit runtime data sparsity to increase efficiency).
Regarding Claim 2, Lai in view of Bourdev and Shumailov teaches the method of claim 1, wherein the second machine learning system is implemented with a neural network (Lai, Para. [0039], “The second device can include a second neural network”).
Regarding Claim 8, Lai teaches a non-transitory computer readable medium comprising computer executable instructions which when executed by a computer cause the computer to perform the method (Para. [0150], “The present disclosure contemplates methods . . . on any machine-readable media . . . using existing computer processors . . . for carrying or having machine-executable instructions”).
The remaining limitations are substantially the same as the limitations of Claim 1, therefore it is rejected under the same rationale.
Regarding Claim 9, the additional elements of the dependent claim are substantially the same as the limitations of Claim 2, therefore it is rejected under the same rationale.
Regarding Claim 14, Lai teaches an apparatus comprising: a memory; and at least one processor, coupled to said memory, and operative to perform operations (Para. [0150], “The present disclosure contemplates methods . . . implemented using . . . machine-readable media [, which] can comprise RAM, ROM, EPROM, EEPROM, or other optical disk storage . . . which can be accessed by a . . . machine with a processor”).
The remaining limitations are substantially the same as the limitations of Claim 1, therefore it is rejected under the same rationale.
Regarding Claim 15, the additional elements of the dependent claim are substantially the same as the limitations of Claim 2, therefore it is rejected under the same rationale.
Claims 3-7, 10-13, and 16-20 are rejected under 35 U.S.C. 103 as being unpatentable over Lai in view of Bourdev, Shumailov, and Park et al. (hereinafter Park) (“Big/Little Deep Neural Network for Ultra Low Power Inference”).
Regarding Claim 3, Lai in view of Bourdev and Shumailov teaches the method of claim 2, wherein the learning of the input transformation function comprises balancing the task loss and the post-activation density of the neural network . . . [at implementation time] (Lai, Para. [0039], “implementing multiple neural networks . . . methods can be implemented to optimize neural network computations for energy savings by splitting the computations across multiple devices . . . the accuracy and one or more corresponding threshold values to determine if the second device is required for further analysis of the input data”; where the “threshold” balances accuracy and energy consumed; Bourdev, Para. [0037], “a task loss, refers to the accuracy of the predicted ML task”, as discussed above, task loss is used to measure accuracy; and Shumailov, Pg. 215, Para. 3, “inputs that lead to less sparse activations will increase the number of operations and the number of memory accesses, and thus energy consumption”, as discussed above, post-activation density is used to measure energy consumed).
The reasons of obviousness have been noted in the rejection of Claim 1 above and remain applicable here.
Lai in view of Bourdev and Shumailov does not teach . . . at inference time based on a specified tradeoff factor . . . .
However, Park teaches . . . at inference time based on a specified tradeoff factor . . . (Pg. 127, Para. 8, “The hyper-parameter [lambda] enables us to make a trade-off between energy consumption and inference accuracy”).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the learning based on minimizing task-specific loss and post-activation density during inferencing, to develop an input transformation function for use by a neural network of Lai as previously modified by Bourdev and Shumailov, with balancing based on a specified tradeoff factor of Park to allow for the development, selection, and use of a multiple machine learning systems to meet the specific needs of a user (Park, Pg. 130, Para. 2, “our current methodology . . . design is to explore all the possible combinations . . . and select the best one which gives the maximum reduction in energy consumption under the given limit of accuracy loss”).
Regarding Claim 4, Lai in view of Bourdev, Shumailov, and Park teach the method of claim 1, further comprising repeating a learning of the input transformation function (Park, Pg. 126, Para. 5, “We assume that each [neural network] is trained by supervised learning”, where the neural network is the input transformation function; Park, Pg. 129, Para. 5, “We give the results of 10 runs”)
at different specified tradeoff factors to generate additional input transformation functions (Park, Figure 7; Park, Pg. 129, Para. 3, “Figure 7 shows the relationship between energy consumption and inference accuracy which is obtained by varying the static threshold from 0 to 1”; Pg. 127, Para. 8, “The hyper-parameter [lambda] enables us to make a trade-off between energy consumption and inference accuracy”; Park, Pg. 127, Equation 4, where “lambda” tradeoff factors are used to calculate “static threshold[s]”, which differentiates the input transformation function neural networks),
wherein the transforming the input data using the learned input transformation function is performed using a selected one of the input transformation functions corresponding to a selected tradeoff factor of the specified tradeoff factors (Park, Pg. 130, Para. 2, “our current methodology . . . design is to explore all the possible combinations . . . and select the best one”; Park, Pg. 130, Figure 7, where users can view results by threshold in order to specify a specific input transformation function to perform the input transformation, which corresponds with a selected tradeoff factor).
The reasons of obviousness have been noted in the rejection of Claim 3 above and remain applicable here.
Regarding Claim 5, Lai in view of Bourdev, Shumailov, and Park teach the method of claim 4, further comprising obtaining the selection of one of the input transformation functions (Park, Pg. 130, Para. 2, “our current methodology . . . design is to explore all the possible combinations of given big and little DNNs and select the best one which gives the maximum reduction in energy consumption under the given limit of accuracy loss”, where selecting a possible combination, which includes the “little DNN” input transformation function, in a manner that “gives” the desired result indicates the selection was obtained).
The reasons of obviousness have been noted in the rejection of Claim 3 above and remain applicable here.
Regarding Claim 6, Lai in view of Bourdev, Shumailov, and Park teach the method of claim 5, wherein the obtained selection is one of: the selected tradeoff factor, an energy level, an energy saving level, an identification of one of the input transformation functions, and an accuracy level (Park, Pg. 130, Para. 2, “As our future work, we will work on design methodologies for the selection of good little DNN (offering large energy reduction while satisfying the given limit of accuracy loss”, where the “little DNN” is the input transformation function and “satisfying the given limit of accuracy loss” indicts the obtained selection is an accuracy level).
The reasons of obviousness have been noted in the rejection of Claim 4 above and remain applicable here.
Regarding Claim 7, Lai in view of Bourdev, Shumailov, and Park teach the method of claim 4, wherein, in the learning step, minimizing the task loss ensures satisfactory accuracy (Lai, Para. [0039], “The first device may use . . . the accuracy . . . to determine whether or not it can make a sufficiently accurate determination without using the second device”, where sufficient accuracy is ensured, which is reasonably understood as minimizing accuracy loss, through use of “the second device[‘s]” neural network when accuracy is insufficient; Bourdev, Para. [0037], “a task loss, refers to the accuracy of the predicted ML task”, as discussed above, task loss is used to measure accuracy)
for the inferencing task (Shumailov, Abstract, “high energy costs of neural network training and inference”)
while minimizing the post-activation density loss ensures that a density of post-activations will be sufficiently small to attain the reduced energy consumption (Lai, Para. [0039], “The systems and methods can be implemented to optimize neural network computations for energy savings by splitting the computations across multiple devices”, where energy consumed in the second device is “optimize[d]”, which is reasonably understood as minimized, which reasonably ensures it, given the inherit tradeoff with accuracy that also must be ensured; Shumailov, Pg. 215, Para. 3, “inputs that lead to less sparse activations will increase the number of operations and the number of memory accesses, and thus energy consumption”, as discussed above, post-activation density is used to measure energy consumed),
and wherein each specified tradeoff factor serves to balance the task loss and the post-activation density loss to enable a tradeoff between the accuracy and the energy consumption (Park Pg. 127, Para. 8, “The hyper-parameter [lambda] enables us to make a trade-off between energy consumption and inference accuracy”, where, as discussed above, the balancing is between task loss and the post-activation density loss).
The reasons of obviousness have been noted in the rejections of Claim 1, for combination with Bourdev and Shumailov, and Claim 4, for combination with Park, above and remain applicable here.
Regarding Claim 10, the additional elements of the dependent claim are substantially the same as the limitations of Claim 3, therefore it is rejected under the same rationale.
Regarding Claim 11, the additional elements of the dependent claim are substantially the same as the limitations of Claim 4, therefore it is rejected under the same rationale.
Regarding Claim 12, the additional elements of the dependent claim are substantially the same as the limitations of Claim 5, therefore it is rejected under the same rationale.
Regarding Claim 13, the additional elements of the dependent claim are substantially the same as the limitations of Claim 6, therefore it is rejected under the same rationale.
Regarding Claim 16, the additional elements of the dependent claim are substantially the same as the limitations of Claim 3, therefore it is rejected under the same rationale.
Regarding Claim 17, the additional elements of the dependent claim are substantially the same as the limitations of Claim 4, therefore it is rejected under the same rationale.
Regarding Claim 18, the additional elements of the dependent claim are substantially the same as the limitations of Claim 5, therefore it is rejected under the same rationale.
Regarding Claim 19, the additional elements of the dependent claim are substantially the same as the limitations of Claim 6, therefore it is rejected under the same rationale.
Regarding Claim 20, the additional elements of the dependent claim are substantially the same as the limitations of Claim 7, therefore it is rejected under the same rationale.
Response to Arguments
Applicant's arguments filed on August 5, 2025 and August 29, 2025 have been fully considered. Each argument is addressed in detail below.
I. Applicant argues the objections to the specification should be withdrawn (Applicant’s Remarks, 08/05/2025, Pg. 6-7, Section I; Applicant’s Remarks 08/29/2025, Pg. 6-7, Section I).
Specifically, Applicant cited aspects of the specification to argue the form and contents of the specification, in regards to Fig. 4, Fig. 6, and reference to Moreira et al., are sufficient as originally presented.
In response to this argument, the objections to the specification have been withdrawn.
II. Applicant argues the rejections of Claims 3-7, 10-13, 16, and 20, under 35 U.S.C. § 112(b), should be withdrawn (Applicant’s Remarks, 08/05/2025, Pg. 8-9, Section II; Applicant’s Remarks 08/29/2025, Pg. 7-9, Section III).
Specifically, Applicant argues the amendments to the claims address all of the issues related to indefiniteness raised in the May 5, 2025 Office Action.
Applicant’s amendments have overcome each and every 112(b) rejection previously set forth in the May 5, 2025 Office Action. As a result, the rejections of Claims 3-7, 10-13, 16, and 20, under 35 U.S.C. § 112(b) have been withdrawn.
III. Applicant argues the rejections of Claims 1-20 as being directed to an abstract idea without significantly more, under 35 U.S.C. § 101, are improper (Applicant’s Remarks, 08/05/2025, Pg. 9-12, Section III; Applicant’s Remarks 08/29/2025, Pg. 9-12, Section III).
Specifically, applicant argues the claimed embodiments reflect technical benefits that are recited in the specification, such as inference-time energy reduction and dynamic adjustment of an energy-accuracy tradeoff, which is an improvement in the technological field of computerized machine learning that amounts to integration into a practical application because it is directed to the application or use of a particular machine. In support of this argument, Applicant cites aspects of the specification, elements of the claims, and MPEP 2106.04(d)(1), while pointing out that the analysis of whether a claim pertains to an improvement to the functioning of a computer or to another technology is without reference to what is well-understood, routine, conventional activity.
According to MPEP 2106.04(d)(1), “[a] claim reciting a judicial exception is not directed to the judicial exception if it also recites additional elements demonstrating that the claim as a whole integrates the exception into a practical application”.
Whereas according to MPEP 2106.05(f), “[t]he recitation of claim limitations that attempt to cover any solution to an identified problem with no restriction on how the result is accomplished and no description of the mechanism for accomplishing the result, does not integrate a judicial exception into a practical application or provide significantly more because this type of recitation is equivalent to the words apply it . . . In contrast, claiming a particular solution to a problem or a particular way to achieve a desired outcome may integrate the judicial exception into a practical application or provide significantly more” (internal quotation marks omitted) (see Electric Power Group, LLC v. Alstom, S.A., 830 F.3d 1350, 1356, 119 USPQ2d 1739, 1743-44 (Fed. Cir. 2016); Intellectual Ventures I v. Symantec, 838 F.3d 1307, 1327, 120 USPQ2d 1353, 1366 (Fed. Cir. 2016); Internet Patents Corp. v. Active Network, Inc., 790 F.3d 1343, 1348, 115 USPQ2d 1414, 1417 (Fed. Cir. 2015); Electric Power, 830 F.3d at 1356, 119 USPQ2d at 1743).
Furthermore, according to MPEP 2106.05(h), “limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application”.
Here, the claims recite the solutions of “minimization of a summation of task loss and a post-activation density loss”, “reconfiguring the post-activation density”, and “reducing the amount of energy consumed for an inferencing task” without claiming a particular way of achieving these desired outcomes. Instead, the claims attempt to cover any mechanism that would achieve these outcomes, while reciting generic computer components like “machine learning system[s]” and a “input transformation function”. Furthermore, the “reducing the amount of energy consumed for an inferencing task” is a natural byproduct and result of the claimed functional language of “reconfiguring the post-activation density”. Therefore, and as discussed in the rejection of Claim 1 under 35 U.S.C. § 101 above, the “reducing the amount of energy consumed for an inferencing task” amounts to generally linking the mental process of “reconfiguring the post-activation density” to the particular technological environment or field of use where the mental process is intended to be used.
As a result, the argument is not persuasive.
IV. Applicant argues the rejections of Claims 1, 2, 8, 9, 14, and 15 as being unpatentable over Lai in view of Bourdev and Shumailov, under 35 U.S.C. § 103, are improper (Applicant’s Remarks, 08/05/2025, Pg. 13-14, Section IV; Applicant’s Remarks 08/29/2025, Pg. 12-14, Section IV).
Specifically, Applicant argues that claims 1, 8, and 14 are patentable because Lai, Bourdev, and Shumailov fail to disclose or suggest, alone or in combination, "learning, using a first machine learning system, an input transformation function that transforms input data for a second machine learning system, the learning using a minimization of a summation of a task loss and a post-activation density loss". Additionally, Applicant argues the dependent claims are patentable at least by virtue of their dependence on independent claims 1, 8, and 14.
According to MPEP 2145 (IV), “One cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references.”
Additionally, according to MPEP 2144.01, “[I]n considering the disclosure of a reference, it is proper to take into account not only specific teachings of the reference but also the inferences which one skilled in the art would reasonably be expected to draw therefrom.” (see also In re Preda, 401 F.2d 825, 826, 159 USPQ 342, 344 (CCPA 1968)).
Furthermore, according to g to 37 CFR 1.111(b), a proper response to an Office action "must be reduced to a writing which distinctly and specifically points out the supposed errors in the examiner’s action.”
Finally, according to MPEP 2143(I), “example rationales that may support a conclusion of obviousness include: . . . Some teaching, suggestion, or motivation in the prior art that would have led one of ordinary skill to modify the prior art reference or to combine prior art reference teachings to arrive at the claimed invention . . . courts have made clear that the teaching, suggestion, or motivation test is flexible and an explicit suggestion to combine the prior art is not necessary. The motivation to combine may be implicit and may be found in the knowledge of one of ordinary skill in the art, or, in some cases, from the nature of the problem to be solved.”
Here, Applicant initially argued against the rejection by pointing out that each of Lai, Bourdev, and Shumailov, considered individually, fail to teach or suggest the element of "learning, using a first machine learning system, an input transformation function that transforms input data for a second machine learning system, the learning using a minimization of a summation of a task loss and a post-activation density loss". However, the rejections of claims 1, 8, and 14 were based on a combination of references to teach this element. As a result, the argument is not persuasive.
Notably, as a part of this initial argument, Applicant states that Shumailov merely states that “inputs that lead to less sparse activations will increase the number of operations and the number of memory accesses, and thus energy consumption” (Pg. 215, Para. 3). This is seemly intended to disprove the assertion that Shumailov teaches “minimizing energy consumption by minimizing post-activation density loss”. While the initial argument as a whole is already not persuasive for the reasons articulated in the preceding paragraph, further discussion on this point is worthwhile for clarity of the record and to expedite prosecution. The position articulated in the May 5, 2025 Office Action and reproduced above is that “sparse activations” are within the broadest reasonable interpretation of activation density and if inputs leading to less sparse activations increase activation density and lead to increased energy consumption, then a person of ordinary skill in the art would understand that inputs leading to more sparse activations decrease activation density and lead to decreased energy consumption. This relationship between decreasing activation density and decreasing energy consumed is inherently required based on the relationship of increasing activation density and increasing energy consumed, or, at the very least, is an inference which one skilled in the art would reasonably be expected to draw. As a result, the argument is not persuasive.
Subsequently, Applicant asserts Lai, Bourdev, and Shumailov, in combination, do not disclose or suggest "learning, using a first machine learning system, an input transformation function that transforms input data for a second machine learning system, the learning using a minimization of a summation of a task loss and a post-activation density loss". The teaching, suggestion, or motivation in the prior art of record that would have led one of ordinary skill to modify or combine the prior art reference to arrive at the claimed invention was articulated in the May 5, 2025 and is reproduced above. However, Applicant does not distinctly or specifically point out any errors in regards to the combination. As a result, the argument is not persuasive.
V. Applicant argues the rejections of Claims 3-7, 10-13, and 16-20 as being unpatentable over Lai in view of Bourdev, Shumailov, and Park, under 35 U.S.C. § 103, are improper (Applicant’s Remarks, 08/05/2025, Pg. 14, Section V; Applicant’s Remarks 08/29/2025, Pg. 14, Section V).
Specifically, Applicant argues the dependent claims are patentable at least by virtue of their dependence on independent claims 1, 8, and 14, which Applicant argues are patentable. This argument is based entirely upon the argued patentability of the independent claims, which as discussed in detail above is not a persuasive argument. As a result, the argument is not persuasive.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MATTHEW BRYCE GOLAN whose telephone number is (571)272-5159. The examiner can normally be reached Monday through Friday, 8:00 AM to 5:00 PM 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, Alexey Shmatov can be reached at (571) 270-3428. 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.
/MATTHEW BRYCE GOLAN/Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights