DETAILED ACTION
This non-final office action is responsive to application 18/413,393 as submitted 16 Jan. 2024.
Claim status is currently pending and under examination for claims 1-20 of which independent claims are 1, 8 and 15.
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
As required by MPEP 609(c), the applicant’s submissions of the Information Disclosure Statement dated 01/16/24 is acknowledged by the examiner and the cited references have been considered in the examination of the claims now pending. As required by MPEP 609 C(2), a copy of the PTOL-1449 initialed and dated by the examiner is attached to the instant office action.
Claim Objections
Claims 5-7, 12-14 and 18-20 are objected to because of the following informalities: the identified claims recite “in an event” which is a contingent limitation that does not positively require the following limitation to be further limiting, e.g. when there is ‘not an event’ such as an else-statement in computer programming, see MPEP 2111.04(II). Appropriate correction is requested.
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 an abstract idea without significantly more. In determining whether the claims are subject matter eligible, the examiner applies guidance set forth under MPEP 2106.
Step 1: Is the claim to a process, machine, manufacture, or composition of matter? Yes—all claims fall within one of the four statutory categories: claims 1-7 are a method/process, claims 8-14 are a computer program product/article of manufacture, and claims 15-20 are a system/machine. As such, all of the claims are to statutory subject matter and the analysis should proceed per MPEP 2106.03.
Step 2A, prong one: Does the claim recite an abstract idea, law of nature or natural phenomenon? Yes—the claims, under the broadest reasonable interpretation, recites an abstract idea. In this case, claims fall within the enumerated grouping of abstract idea comprising “Mental Processes” and/or “Mathematical Concepts”, but for the recitation of generic computer components. In particular, claims recite:
“recognizing that a change in data used […] occurs”
(Mental observation, e.g. detecting data difference)
“computing an expected model change for each of the models in accordance with the change in the data” (Mental judgment with evaluation, or math calculating/computing)
“predicting a resource consumption level associated with […] the model” (Mental estimate of needed ‘resource’ which could be time, money, expertise, or data for model/hypothesis)
“computing, over a set of models, an optimal model […] score for each of the models based on the expected model change and the resource consumption level” (Mathematical calculation/computing, e.g. [0071] “The score is computed as…” Equation 3)
“comparing the optimal model retraining score with a threshold” (Mental comparison as evaluation and/or math relationship using less than/greater than operands, e.g. ≥, <, argmax function or the like)
When read in light of the specification, Fig 3 clearly shows math detailed by the specification for a score that is ‘computed’ by calculation of equation 3 [0071], and describes end users including customers [0048] whom are humans capable of performing mental tasks such as recognizing, estimating and comparing. Therefore, the claims recite at least mental processes and/or mathematical concepts as abstract ideas enumerated under MPEP 2106.04(a)(2)(I,III).
Step 2A, prong two: Does the claim recite additional elements that integrate the judicial exception into a practical application? No—a practical application is not integrated by the judicial exception because the additional elements are as follows:
“computer-implemented” MPEP 2106.05(f) merely uses a computer as a tool to perform an abstract idea, recited at a high level of generality.
“executing automatic model retraining based on a CO2-based model retraining score” and “executing an automatic model retraining for each model for which the optimal model retraining score exceeds the threshold” MPEP 2106.05(h) generally linking the use of the judicial exception to a particular technological environment or field of use.
Balance of the claim employs a computer for executing automatic model retraining. The execution of retraining entails scoring that generally relates to CO2 at level of preamble only, however the retrained model does not output anything to impart application, much less convey the practical nature thereof. It simply performs repetitive training optimization which does not meaningfully limit the claim beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is no more than a drafting effort to monopolize the exception. Further, as per MPEP 2106.04(a)(2)(III) “A claim that requires a computer may still recite a mental process.” Accordingly, the claims remain drawn to the abstract idea and the additional elements are insufficient to integrate the judicial exception into a practical application.
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception? No—the claims do not include additional elements that amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional elements are identified with respect to MPEP 2106.05 and do not demonstrate an inventive concept. Particularly, the additional elements are as follows:
“computer-implemented” MPEP 2106.05(f) merely uses a computer as a tool to perform an abstract idea, recited at a high level of generality. Particularly, a general purpose computer does not qualify as a particular machine under MPEP 2106.05(b).
“executing automatic model retraining based on a CO2-based model retraining score” and “executing an automatic model retraining for each model for which the optimal model retraining score exceeds the threshold” MPEP 2106.05(h) generally linking the use of the judicial exception to a particular technological environment or field of use. Particularly, the elements do not provide a meaningful limitation through a technical solution in a manner that elevates the claim as a whole beyond established techniques under MPEP 2106.05(e).
Significantly more is not satisfied for at least the reasons identified above. If the claim language provides only a result-oriented solution, with insufficient detail for how a computer accomplishes it, then the claims do contain an inventive concept. Taken alone, their additional elements do not amount to significantly more than the above-identified judicial exception. Looking at the claim as a whole with limitations as an ordered combination adds nothing that is not already present when looking at the elements taken individually. The collective functions are merely conventional computer implementation. Accordingly, claim 1 is found ineligible for patent under §101.
Independent claim 8 recites limitations similar to claim 1 and further recites “computer program product comprising one or more computer readable storage media having computer readable program code collectively stored on the one or more computer readable storage media, the computer readable program code being executed by a processor of a computer system” which are additional elements that fall under MPEP 2106.05(f) mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Particularly, the additional elements are recited at a high level of generality and do not qualify as a particular machine under MPEP 2106.05(b). Thus, the claim remains drawn to the abstract idea and the additional elements do not integrate the abstract idea into a practical application or amount to significantly more.
Independent claim 15 recites limitations similar to claim 1 and further recites “A computing system comprising: a processor; a memory coupled to the processor; and one or more computer readable storage media coupled to the processor, the one or more computer readable storage media collectively containing instructions that are executed by the processor via the memory” which are additional elements that fall under MPEP 2106.05(f) mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Particularly, the additional elements are recited at a high level of generality and do not qualify as a particular machine under MPEP 2106.05(b). Thus, the claim remains drawn to the abstract idea and the additional elements do not integrate the abstract idea into a practical application or amount to significantly more.
Dependent claims 2, 9 and 16 disclose where the resource consumption level comprises a level of carbon emissions. This is considered part of the abstract idea that embellishes the predicted resource e.g. estimating carbon footprint. The ‘level’ does not impart a particular unit of measurement such as kilograms or tons, and for carbon emission to emit, it must necessarily emit some nonzero amount. The functionality is similar to projecting that more carbon is emitted by air-travel than traveling by car, or that monetary capital resource required to fill a car’s gas tank with fuel may burn resulting in some CO2 out the exhaust pipe. There are no additional elements.
Dependent claims 3 and 10 disclose wherein the level of carbon emissions comprises total carbon emissions associated with CPU usage and activation and usage of server racks and storage units. The usage of CPU, activation, server racks and storage units are considered additional elements under the apply-it (i.e. usage) consideration of MPEP 2106.05(f). A CPU (as opposed to GPU or TPU for AI/ML) and storage units are highly generic, server racks can be mainframes from by-gone era, and no detail is given about the activation’s technical nature. However, the combination of elements might suggest a data center’s collective/total carbon emissions. This is an established field of research that generally links the use of the judicial exception to a particular technological environment or field of use under MPEP 2106.05(g). Supplemental evidence is provided as per Faiz LLMcarbon (below) at Fig. 1. As such, the additional elements are insufficient to integrate the judicial exception into a practical application or amount to significantly more.
Dependent claims 4 and 11 disclose wherein the computing of the score comprises analyzing carbon emissions tradeoff with retrained model performance. The limitation is considered part of the abstract idea such that analyzing tradeoffs can be mental evaluation and/or mathematical relationships. For example, proportionality relating compute power or training time (as a proxy for carbon emission) to accuracy as a model performance evaluation. There are no additional elements.
Dependent claims 5, 12 and 18 disclose evaluating resource consumption changes and balancing resource allocation based thereon. This is considered part of the abstract idea being mental processes to include mental evaluation and judgment to assign/allocate resources in a suitable manner. For example, assigning an increasingly complex model to a high performing asset or divide-and-conquer approach. There are no additional elements.
Dependent claims 6, 13 and 19 disclose evaluating time intervals between model retraining. This is considered part of the abstract idea being mental evaluation. For example, daily or weekly intervals set according to calendar schedule. There are no additional elements.
Dependent claim 7, 14 and 20 disclose evaluating business costs of model retraining. This is considered part of the abstract idea being mental evaluation. For example, revenue profits or time-cost evaluated by business entity. There are no additional elements.
Dependent claim 17 discloses limitations of claims 3-4 which are already addressed as above.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-4, 8-11 and 15-17 are rejected under 35 U.S.C. 103 as being unpatentable over:
Das et al., US PG Pub No 2024/0232698A1 hereinafter Das, in view of
Faiz et al., “LLMCarbon: Modeling the End-to-End Carbon Footprint of Large Language Models” hereinafter Faiz (arXiv: 2309.14393v1).
With respect to claim 1, Das teaches:
A computer-implemented method for executing automatic model retraining based on a CO2-based model retraining score, the computer-implemented method {Das Fig 4 method, implemented by computer Figs 1, 8 and introduced [0012-13] “retraining monitoring system collects contextual data… The contextual data along with parameters associated with the components such as but not limited to, energy demand, energy efficiency, carbon emission” emphasis carbon emission, similar at [0019-20], scoring thereof may comprise Eq.1 [0025] using rewards Eqs. 4-6,7 [0030,36]} comprising:
recognizing that a change in data used to train a set of models occurs {Das discloses [0030] “testing for data shift/data drift within the current training data… retraining the ML models” see e.g. [0026-30] feature data over sequence of states and sampling, also Figs 1:140, 7:706};
computing an expected model change for each of the models in accordance with the change in the data {Das [0014] “changing a current retraining platform to another model retraining platform” per Fig 4:410,414 thus the change for models is reasonably expected to be on different retraining platform, [0033-34] “change in the platform or the retraining architecture can be made based on whether the sustainability…meets certain predefined criteria” because “Inefficient location management can lead to excess carbon emissions” similar at [0021], see model update [0025]};
predicting a resource consumption level associated with retraining each of the model {Das [0014-15] “energy consumption of processing may be used in the reward” and “reward value is estimated from the contextual information” where estimating is predicting, see e.g. [0023] “estimating the sustainability of the retraining platforms from the relevant metadata 252 which can include without limitation, the processing duration, and energy source of the data center. The reward is based on the delay of communication and energy consumption” similar [0032] “sustainability estimations of the components such as but not limited to, energy demand, energy efficiency, carbon emission” see also [0035-41], and [0015,25] estimation/prediction may use a gradient monte carlo algorithm};
computing, over a set of models, an optimal model retraining score for each of the models based on the expected model change and the resource consumption level {Das discloses [0017] “reward optimization” scores thereof comprise Eqs. 4-6,7 [0030,36] reward-R plugs into Eq.1 [0025] describes reinforcement learning with multiple policies, policy-based models are employed for the retraining architecture/platform, e.g. [0015] “modeled as multiple policies” policies with rewards comprise energy source which consumes energy for retraining models subject to drift, see [0029-32] “maximizes the reward… each of the ML models 140 have fallen and the model require retraining” see Figs 4-5. Additional example at [0060-66,45]};
comparing the optimal model retraining score with a threshold {Das [0015] “comparisons of the parameter values with the predetermined threshold values… retraining ML models” where the parameter values are based on reward values in feedback loop Fig 5:508-10 and/or [0026]. Notably [0054-55] “if dn>=Predefined threshold value, then… Retrain with a probability of 0.9” and/or [0033-34] “retraining architecture meets certain predetermined criteria… sustainability criteria” similar at [0030] “criteria, e.g., predefined threshold values”}; and
executing an automatic model retraining for each model for which the optimal model retraining score exceeds the threshold {Das [0015] “executing an action… Based on comparisons of the parameter values with the predetermined threshold values, different actions can be executed such as… retraining the ML models” again at [0030] “action provider 106 applies certain criteria, e.g., predefined threshold values to the feature set to select and provide an action… actions that are implemented can include retraining the ML models” see Fig 6:608, [0054-55], [0032], [0019-20] “re-train the ML models”}.
However, Das does not explicitly state that the carbon emission has two oxygen molecules (CO2) which is detailed by Faiz.
Particularly, Faiz [P.5-6 Sect. 4.6-4.8] Eqs.10-13 details CO2eq of carbon model for LLM, Fig 1 noting the LLM as a MoE mixture-of-experts (expert models) specifying number of experts and loss [P.4].
Faiz is directed to carbon emission footprint and energy consumption of trained models thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to specify CO2 model per Faiz in combination to arrive at the invention as claimed as applying known techniques to known methods ready for improvement to yield predictable results using “industry standard metric for evaluating a data center’s energy efficiency” [P.5 Sect. 4.6 ¶2] and/or a motivation “explore the trade-off between test loss and carbon footprint when designing new LLMs… encourages everyday users to adopt practices that mitigate LLM carbon footprints” [P.1 ¶2] and with benefit of “estimating the carbon impact of emerging LLMs even before their training… accurately predict the carbon footprint of both dense and MoE LLMs during their training” [Abst], [P.2 ¶2].
With respect to claim 2, the combination of Das and Faiz teaches the computer-implemented method according to claim 1, wherein the resource consumption level comprises
a level of carbon emissions {Das [0020,32] “carbon emissions” at least some level is requisite. Also see Faiz [P.5-6 Sect. 4.7] Table 3 “kgCO2/cm2” as “Carbon emitted Per unit Area (CPA)” where kgCO2 quantifies a level as kilograms of carbon dioxide, similar at Table 5 [P.8] kgCO2eq (eq denotes equivalent)}.
A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to specify a level of carbon emissions per Faiz for Das’s carbon emissions as obvious to try in choosing from a finite number of identified, predictable solutions for measuring carbon emissions, with a reasonable expectation of success.
With respect to claim 3, the combination of Das and Faiz teaches the computer-implemented method according to claim 2, wherein the level of carbon emissions comprises
total carbon emissions associated with central processing unit (CPU) usage and activation and usage of server racks and storage units {Faiz [P.6 Sect. 4.8] “total carbon footprint (CO2eq) resulting from LLM processing is determined using Equation 13” detailed [P.5 Sect. 4.6] Eq.13 summed terms decompose into Eqs.10-12 where operational carbon model comprises “CPUs, LLM computing devices, memories” memories are storage units, cont’d to comprise “all auxiliary components like cooling, to the energy consumed solely by the computing hardware within the data center” where data centers employ server racks as [P.8 ¶1] “64 8-GPU servers” Tbls 5,3, and activation is interpreted to comprise inference of the LLM [P.7 ¶2], [P.4 Sect. 4.4] inference modeled by flop count, Fig 1 shows total CO2eq as a result of both operational CO2eq under LLM arch config and embodied CO2eq under hardware config}.
A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to use total carbon footprint for emissions per Faiz in combination for a motivation that “overall carbon footprint of the LLM is then computing by summing” [P.3 Last¶] and for an “end-to-end carbon footprint modeling tool for… carbon emissions during training, inference, experimentation, and storage processes” [P.9 Conc.], [P.2 ¶2]. This conveys “the entire life-cycle carbon footprint of the LLM” [P.8 ¶2].
With respect to claim 4, the combination of Das and Faiz teaches the computer-implemented method according to claim 2, wherein the computing of the optimal model retraining score comprises
analyzing carbon emissions tradeoffs with retrained model performance {Das [0021,20-22] “trade-off analyzer 106… if the accuracy of the retrained model is higher than the accuracy of the prior models, then…further training” where accuracy is a performance measurement. Additionally see Faiz Fig 8 “trade-off between training carbon footprint and test loss”}.
A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to include the trade-off analysis of Faiz for the trade-off analyzer of Das in combination for a motivation of “enabling efficient exploration of the design space by considering the trade-off between carbon footprint and test loss” [P.9 Conc.], [P.1 ¶2].
With respect to claim 8, the rejection of claim 1 is incorporated. The difference in scope being a computer program product comprising computer readable media storing program code executed by processor of computer system to perform limitations of the method claim 1. Das discloses [0069-70] “software or machine-readable instructions stored on a non-transitory processor-readable storage medium and executed by one or more processors” shown Fig 8. The remainder of this claim is rejected for the same rationale as claim 1.
With respect to claim 9, the combination of Das and Faiz teaches the computer program product according to claim 8, and further teaches the limitation of claim 2. Therefore, the rejection of claim 2 with equal motivation is applied to claim 9.
With respect to claim 10, the combination of Das and Faiz teaches the computer program product according to claim 9, and further teaches the limitation of claim 3. Therefore, the rejection of claim 3 with equal motivation is applied to claim 10.
With respect to claim 11, the combination of Das and Faiz teaches the computer program product according to claim 9, and further teaches the limitation of claim 4. Therefore, the rejection of claim 4 with equal motivation is applied to claim 11.
With respect to claim 15, the rejection of claim 1 is incorporated. The difference in scope being a computing system comprising processor coupled to memory and computer readable storage media containing instructions executed by processor via memory to implement the limitations of claim 1. Das discloses [0068-70] “computer system 800 includes processor(s) 802… non-volatile memory, or a volatile medium such as RAM” being “operatively coupled to a bus 808. The processor-readable medium 806 may be any suitable medium that participates in providing instructions to the processor(s) 802” Fig 8. The remainder of this claim is rejected for the same rationale as claim 1.
With respect to claim 16, the combination of Das and Faiz teaches the computing system according to claim 15, and further teaches the limitation of claim 2. Therefore, the rejection of claim 2 with equal motivation is applied to claim 16.
With respect to claim 17, the combination of Das and Faiz teaches the computing system according to claim 16, and further teaches the limitation of claims 3-4. Therefore, the rejection of claims 3-4 with equal motivation is applied to claim 17.
Claims 5, 12 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Das and Faiz in view of
Hess et al., US PG Pub No 2024/0250529A1 hereinafter Hess.
With respect to claim 5, the combination of Das and Faiz teaches the computer-implemented method according to claim 1, wherein in an event the automatic model retraining is executed, the computer-implemented method further comprises:
evaluating resource consumption changes of retrained models {Das discloses [0036] “The energy consumption of processing can be further based on two factors—the duration of the processing and the energy source…energy source of the data center can have different values” thus different energy sources are changes in consumption of energy resources, cont’d [0041] “energy source of the retraining platform” Fig 4:410,414}; and
However, Das and Faiz do not teach “balancing resource allocation” which is disclosed by Hess:
balancing resource allocation based on the resource consumption changes {Hess per [0106] “Load balancers help determine the most efficient distribution of tasks to optimize resource utilization… allocate network resources” for model trained to predict power consumption [0104] Fig 1, see e.g. [0083,82] “prediction energy consumption and/or carbon related emissions” and [0116] “migration between datacenters”}.
Hess is directed to carbon emission predictions with trained models thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to balance load for allocating resources per Hess in combination to arrive at the invention as claimed for a motivation to “help determine the most efficient distribution of tasks to optimize resource utilization” [0106] which may provide a benefit of “better power consumption” [0026], and because “Along with the growing environmental awareness and legal regulations, many entities are obligated to reduce carbon dioxide emissions” [0031].
With respect to claim 12, the combination of Das and Faiz teaches the computer program product according to claim 8, and further combination with Hess teaches the limitation of claim 5. Therefore, the rejection of claim 5 with equal motivation is applied to claim 12.
With respect to claim 18, the combination of Das and Faiz teaches the computing system according to claim 15, and further combination with Hess teaches the limitation of claim 5. Therefore, the rejection of claim 5 with equal motivation is applied to claim 18.
Claims 6, 13 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Das and Faiz in view of
Hughes et al., US PG Pub No 2023/0385836A1 hereinafter Hughes.
With respect to claim 6, the combination of Das and Faiz teaches the computer-implemented method according to claim 1, wherein in an event the automatic model retraining is executed. Hughes teaches the computer-implemented method further comprises
evaluating time intervals between model retraining {Hughes [0043] “time intervals between retrained ML models” and [0042] “the retrained and reweighted model can be evaluated” e.g. shown Fig 1:105,115 or 3:320,330, described [0030,38] “daily retraining… set a time interval (e.g., daily), gather all data for updates, retrain a machine learning (ML) model, ensure that the retrained model meets organizational thresholds for performance”. Also note Das [0038-39] duration of a process/action, where action is retraining [0015,30]}.
Hughes is directed to model retraining thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to specify daily retraining per Hughes for Das’s duration to arrive at the invention as claimed for a motivation “Advantages to frequent (e.g., daily) model retraining and rule optimization include near-real time recognition/prediction of emerging patterns in recent historical data” [0041] and/or “goal of daily model retraining, a model can be created each day… effectively produce unbiased results with respect to more recent training” [0053].
With respect to claim 13, the combination of Das and Faiz teaches the computer program product according to claim 8, and further combination with Hughes teaches the limitation of claim 6. Therefore, the rejection of claim 6 with equal motivation is applied to claim 13.
With respect to claim 19, the combination of Das and Faiz teaches the computing system according to claim 15, and further combination with Hughes teaches the limitation of claim 6. Therefore, the rejection of claim 6 with equal motivation is applied to claim 19.
Claims 7, 14 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Das and Faiz in view of
Luo et al., US PG Pub No 2018/0197087A1 hereinafter Luo.
With respect to claim 7, the combination of Das and Faiz teaches the computer-implemented method according to claim 1, wherein in an event the automatic model retraining is executed. Luo teaches the computer-implemented method further comprises
evaluating business costs of model retraining {Luo Fig 1:112 Model Retraining, [0074-79] “financial costs associated with retraining a current model… threshold can correspond to a particular retraining cost amount (e.g., $100,000), while the impact metric can correspond to the profit loss indicator”, e.g. [0046] ”cost to retrain the current model… business cost/impact such as profit loss”}.
Luo is directed to model retraining thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to consider financial cost as a business cost based on retraining models per Luo to arrive at the invention as claimed for a motivation “retraining trigger that has a relatively high impact… $10,000 impact/reduction in business profitability” [0075-77].
With respect to claim 14, the combination of Das and Faiz teaches the computer program product according to claim 8, and further combination with Luo teaches the limitation of claim 7. Therefore, the rejection of claim 7 with equal motivation is applied to claim 14.
With respect to claim 20, the combination of Das and Faiz teaches the computing system according to claim 15, and further combination with Luo teaches the limitation of claim 7. Therefore, the rejection of claim 7 with equal motivation is applied to claim 20.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Luccioni et Hernandez-Garcia, “Counting Carbon: A Survey of Factors Influencing the Emissions of Machine Learning” arXiv: 2302.08476v1, [P.2] Eq.1 “amount of CO2eq emitted during model training can be decomposed…” Eq.1
Eilam et al., “Towards a Methodology and Framework for AI Sustainability Metrics” IBM authors-differ from application (inventorship?), see [Sect. 2.3] “carbon cost of re-training at that interval… cost of re-training which is use-case specific”
Radovanovic et al., “Carbon-Aware Computing for Datacenters” arXiv: 2106.11750v1 Google authors, [P.6 ¶2] “models are retrained and evaluated daily”
Tosi et al., US PG Pub No 2025/0139501A1 Mind Foundry, carbon footprint discloses retrain
Aurongzeb et al., US PG Pub No 2024/0354172A1 Dell, data center carbon footprint
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Chase P Hinckley whose telephone number is (571)272-7935. The examiner can normally be reached M-F 9:00 - 5:00.
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, Miranda M. Huang can be reached at 571-270-7092. 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.
/CHASE P. HINCKLEY/Examiner, Art Unit 2124
Prosecution Insights