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 .
Response to Arguments
Applicant’s arguments with respect to claims 1-29 have been considered but are moot because the arguments do not apply to any of the references being used in the current rejection.
Claim Rejections - 35 USC § 102
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 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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless -
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1, 7-13, 21-23, and 29 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Marzorati et al. (US 2023/0236871, hereinafter Marzorati).
Regarding claim 1, Marzorati discloses
A method comprising (fig. 1-12):
using one or more machine learning models to output one or more predicted actions (paragraphs [0005], [0032]-[0039], [0066]-[0069]: training and using predictive ML models (CE predictive models and a global predictive model) on utilization/telemetry data to predict future utilization and to produce action decisions to reprovision environments) that are used to modify at least one current state of a computing system (paragraphs [0005], [6006]-[6018], [1106], [1302]: providing an action decision to reprovision and sending provisioning data to modify infrastructure/virtualization configuration) using first inputs to the one or more machine learning models based at least in part on values of at least one operating parameter of the computing system (paragraphs [0032]-[0039], [0046]: obtaining and iteratively using operating/utilization parameters (infrastructure utilization, network utilization, CPU, memory, throughput, requests/sec, etc.) as model inputs and trains CE predictive models on those parameters), second inputs to the one or more machine learning models based at least in part on values of at least one system objective (paragraphs [0041], [0062]-[0066], [0086]: storing and using SLA/service-objective data (SLA repository) and explicitly uses SLA/availability targets and other objective parameters as inputs to the decision process when querying models and determining provisioning), and at least one desired state of the computing system determined at least in part by the at least one system objective (paragraphs [0066]-[0069], [6006]-[6016]: querying global/CE predictive models with target utilization and SLA availability to derive infrastructure/virtualization configurations predicted to meet the SLA) (Note: ML pipeline ingests operating telemetry (i.e., first inputs) and SLA/objective data (i.e., second inputs), predicts future utilization/needs, queries models for configurations that meet the desired SLA state (i.e., desired state), and produces “action decisions” / provisioning data to reconfigure the environment); and
causing an application, which receives the one or more predicted actions output from the one or more machine learning models, to implement the one or more predicted actions to modify the current state of the computing system to a different state of the computing system (paragraphs [0102], [6018], [1106], [1302]: sending the provisioning data/actions from the orchestrator to manager/agent components that carry out reprovisioning (deploying nodes/VMs, adjusting allocations)).
Regarding claim 7, Marzorati discloses
further comprising: performing at least one workload on the computing system (paragraph [0027]: edge network disposed computing environments … can run a set of common software applications,” including “image recognition … AI support … VNF software applications,” often “requiring low latency performance; paragraph [0120]: “edge computing environments can dynamically be updated … to support a specific number of edge computing requests as per a defined service level agreement), the at least one system objective being associated with the at least one workload (paragraph [0041]: Orchestrator 110 in service level agreement (SLA) repository 2122 can store current SLA data for each of the respective computing environments).
Regarding claim 8, Marzorati discloses
further comprising: occasionally repeating selecting the one or more predicted actions, and causing the application to implement the one or more predicted actions (Fig. 6; paragraph [0081]: On completion of block 6002, orchestrator 110 can proceed to block 6004. At block 6004, orchestrator 110 can perform discovering of a growth trend of utilization data with respect to a target computing environment based on an analysis of historical data; paragraph [0090]: Based on the return data returned at block 6006, orchestrator 110 can proceed to block 6008. At block 6008, orchestrator 110 can virtually reprovision and digitally adjust a configuration for the target computing environment currently being evaluated for provisioning in dependence on the returned adjustment data returned at block 6006).
Regarding claim 9, Marzorati discloses
further comprising: using at least one of machine learning or artificial intelligence (paragraph [0120] Embodiments herein can provide an artificial intelligence (AI) based method and system by which edge computing environments can dynamically be updated with new configurations including new sensor configurations to gather data and computing resources to process the gathered data to support a specific number of edge computing requests as per a defined service level agreement) to obtain the at least one current state of the computing system (paragraphs [0066]-[0069]: At training block 1103, orchestrator 110 can perform training of various predictive models … CE predictive model 3002 can be trained with iterations of training data … trained as described, CE predictive model 3002 can be configured to respond to query data).
Regarding claim 10, Marzorati discloses
further comprising: wherein the one or more predicted actions comprise at least one of modifying a number of workloads being performed by the computing system or modifying hardware resources of the computing system (paragraph [0027]: Edge network disposed computing environments … can run a set of common software applications … Certain software applications … require low latency performance demanded by UE device clients being serviced by the running of the software applications; paragraph [0102]: orchestrator 110 can send provisioning data to the target computing environment of computing environments 120A-120Z and that target computing environment at provisioning block 1302 can be reprovisioned according to the provision data, e.g., by adjustment of a count of computing nodes, adjustment of computing node types, adjustment of virtualization parameter values expressed, e.g., by increases in numbers (a count) of virtual machines, adjustment of virtual machine types, and/or adjustment of virtual machine resource provisioning).
Regarding claim 11, Marzorati discloses
A system comprising: one or more hardware resources; a processing environment comprising at least a portion of the one or more hardware resources; and one or more circuits to (Fig. 10):
obtain values of one or more parameters as one or more workloads are performed by the processing environment in a current state; use one or more machine learning models (paragraphs [0005], [0032]-[0039], [0066]-[0069]: training and using predictive ML models (CE predictive models and a global predictive model) on utilization/telemetry data to predict future utilization and to produce action decisions to reprovision environments) to select at least one predicted action (paragraphs [0005], [6006]-[6018], [1106], [1302]: providing an action decision to reprovision and sending provisioning data to modify infrastructure/virtualization configuration) based at least in part on first inputs (paragraphs [0032]-[0039], [0046]: obtaining and iteratively using operating/utilization parameters (infrastructure utilization, network utilization, CPU, memory, throughput, requests/sec, etc.) as model inputs and trains CE predictive models on those parameters) and second inputs (paragraphs [0041], [0062]-[0066], [0086]: storing and using SLA/service-objective data (SLA repository) and explicitly uses SLA/availability targets and other objective parameters as inputs to the decision process when querying models and determining provisioning), the first inputs to be based at least in part on the values of the one or more parameters (paragraphs [0032]-[0039], [0046]: obtaining and iteratively using operating/utilization parameters (infrastructure utilization, network utilization, CPU, memory, throughput, requests/sec, etc.) as model inputs and trains CE predictive models on those parameters), the second inputs to be based at least in part on values of one or more objectives associated with the one or more workloads (paragraphs [0041], [0062]-[0066], [0086]: storing and using SLA/service-objective data (SLA repository) and explicitly uses SLA/availability targets and other objective parameters as inputs to the decision process when querying models and determining provisioning), selecting the at least one predicted action to comprise:
predicting one or more predicted actions (paragraphs [0088]: At block 6006, orchestrator 110 on querying global CEs predictive model 4002 with the described query data … can obtain return data that specifies infrastructure parameter values and/or virtualization parameter values that are predicted to support the referenced utilization parameter values including availability parameter values input as query data at a future time period; paragraph [0097]: At block 6014, orchestrator 110 can run the identified simulation scenarios using a computing environment predictive model 3002 associated to the identified matching computing environment … Running a simulation scenario, block 6014 can include applying as query data to the identified CE predictive model 3002 … query data that comprises computing environment characterizing parameter values associated with the boundary condition; Note: Marzorati teaches applying each candidate action (candidate reprovisioning) to predictive models to obtain predicted future states (simulated outcomes) for that candidate) based at least in part on the first inputs (paragraphs [0032]-[0039], [0046]: obtaining and iteratively using operating/utilization parameters (infrastructure utilization, network utilization, CPU, memory, throughput, requests/sec, etc.) as model inputs and trains CE predictive models on those parameters) and the second inputs paragraphs [0041], [0062]-[0066], [0086]: storing and using SLA/service-objective data (SLA repository) and explicitly uses SLA/availability targets and other objective parameters as inputs to the decision process when querying models and determining provisioning);
predicting sets of one or more different future states of the processing environment if the one or more predicted actions are taken with respect to the processing environment (paragraph [0085]: At block 6006, orchestrator 110 can query global CEs predictive model 4002 … for return of target computing environment infrastructure and/or virtualization configuration adjustment data; paragraph [0095]: On completion of block 6010, orchestrator 110 can proceed to block 6012. At block 6012, orchestrator 110 can discover simulation scenarios for testing the virtually reprovisioned target computing environment virtually reprovisioned with returned data returned at block 6006; Note: Marzorati teaches generating multiple candidate provisioning adjustments (i.e., a plurality of candidate actions) and using predictive models to evaluate them); and
determining a selected set of the sets that more closely matches at least one desired state of the processing environment than at least one other of the sets, the selected set having been predicted for the at least one predicted action of the one or more predicted actions (paragraph [0090]: Based on the return data returned at block 6006, orchestrator 110 can proceed to block 6008. At block 6008, orchestrator 110 can virtually reprovision and digitally adjust a configuration for the target computing environment); and
perform the at least one predicted action to modify the current state of the processing environment to a different state of the processing environment (paragraphs [0102], [6018], [1106], [1302]: sending the provisioning data/actions from the orchestrator to manager/agent components that carry out reprovisioning (deploying nodes/VMs, adjusting allocations)).
Regarding claim 12, Marzorati discloses
wherein the one or more circuits are to instruct the processing environment to perform the one or more workloads (Fig. 10; paragraph [0165]: program processes may include routines, programs, objects, components, logic, data structures, and so on that perform particular tasks or implement particular abstract data types).
Regarding claims 13, Marzorati discloses
wherein the at least one predicted action comprises at least one of modifying the one or more workloads being performed by the processing environment or modifying the portion of the one or more hardware resources of the processing environment (paragraph [0027]: Edge network disposed computing environments … can run a set of common software applications … Certain software applications … require low latency performance demanded by UE device clients being serviced by the running of the software applications; paragraph [0102]: orchestrator 110 can send provisioning data to the target computing environment of computing environments 120A-120Z and that target computing environment at provisioning block 1302 can be reprovisioned according to the provision data, e.g., by adjustment of a count of computing nodes, adjustment of computing node types, adjustment of virtualization parameter values expressed, e.g., by increases in numbers (a count) of virtual machines, adjustment of virtual machine types, and/or adjustment of virtual machine resource provisioning).
Regarding claims 21 and 29, Marzorati discloses
wherein the system is comprised in at least one of:
a control system for an autonomous or semi-autonomous machine;
a perception system for an autonomous or semi-autonomous machine;
a first system for performing simulation operations (paragraph [0055]: Orchestrator 110 performing examining process 111 can perform discovering simulation scenarios using historical data stored in computing environment parameters history area 2121);
a second system for performing deep learning operations;
a third system implemented using an edge device;
a fourth system implemented using a robot;
a fifth system incorporating one or more virtual machines (VMs);
a sixth system implemented at least partially in a data center;
a seventh system for performing digital twin operations;
an eighth system for performing light transport simulation;
a nineth system for performing collaborative content creation for 3D assets;
a tenth system for performing conversational Artificial Intelligence operations;
an eleventh system for generating synthetic data;
a twelfth system for implementing a web-hosted service for detecting program workload inefficiencies;
an application as an application programming interface (“API”);
a thirteenth system implemented at least partially using cloud computing resources; or
a fourteenth system for presenting one or more of virtual reality content, augmented reality content, or mixed reality content.
Regarding claim 22, Marzorati discloses
A processor comprising (fig. 1-12): one or more circuits to (Fig. 10):
obtain values of one or more parameters as one or more workloads are performed by a processing environment in a current state; use one or more machine learning models to generate at least one predicted action (paragraphs [0005], [0032]-[0039], [0066]-[0069]: training and using predictive ML models (CE predictive models and a global predictive model) on utilization/telemetry data to predict future utilization and to produce action decisions to reprovision environments) based on first inputs and second inputs, the first inputs to be based at least in part on the values of the one or more parameters (paragraphs [0032]-[0039], [0046]: obtaining and iteratively using operating/utilization parameters (infrastructure utilization, network utilization, CPU, memory, throughput, requests/sec, etc.) as model inputs and trains CE predictive models on those parameters), the second inputs to be based at least in part on values of one or more objectives associated with the one or more workloads (paragraphs [0041], [0062]-[0066], [0086]: storing and using SLA/service-objective data (SLA repository) and explicitly uses SLA/availability targets and other objective parameters as inputs to the decision process when querying models and determining provisioning), generating the at least one predicted action to comprise (paragraphs [0088]: At block 6006, orchestrator 110 on querying global CEs predictive model 4002 with the described query data … can obtain return data that specifies infrastructure parameter values and/or virtualization parameter values that are predicted to support the referenced utilization parameter values including availability parameter values input as query data at a future time period; paragraph [0097]: At block 6014, orchestrator 110 can run the identified simulation scenarios using a computing environment predictive model 3002 associated to the identified matching computing environment … Running a simulation scenario, block 6014 can include applying as query data to the identified CE predictive model 3002 … query data that comprises computing environment characterizing parameter values associated with the boundary condition; Note: Marzorati teaches applying each candidate action (candidate reprovisioning) to predictive models to obtain predicted future states (simulated outcomes) for that candidate):
using the first inputs (paragraphs [0032]-[0039], [0046]: obtaining and iteratively using operating/utilization parameters (infrastructure utilization, network utilization, CPU, memory, throughput, requests/sec, etc.) as model inputs and trains CE predictive models on those parameters) and the second inputs (paragraphs [0041], [0062]-[0066], [0086]: storing and using SLA/service-objective data (SLA repository) and explicitly uses SLA/availability targets and other objective parameters as inputs to the decision process when querying models and determining provisioning) to predict one or more predicted actions (paragraphs [0088]: At block 6006, orchestrator 110 on querying global CEs predictive model 4002 with the described query data … can obtain return data that specifies infrastructure parameter values and/or virtualization parameter values that are predicted to support the referenced utilization parameter values including availability parameter values input as query data at a future time period; paragraph [0097]: At block 6014, orchestrator 110 can run the identified simulation scenarios using a computing environment predictive model 3002 associated to the identified matching computing environment … Running a simulation scenario, block 6014 can include applying as query data to the identified CE predictive model 3002 … query data that comprises computing environment characterizing parameter values associated with the boundary condition; Note: Marzorati teaches applying each candidate action (candidate reprovisioning) to predictive models to obtain predicted future states (simulated outcomes) for that candidate); and
selecting a selected set from sets of one or more different future states predicted using the one or more predicted actions, the selected set to more closely match at least one desired state of the processing environment than at least one other of the sets, the selected set corresponding to the at least one predicted action (paragraph [0090]: Based on the return data returned at block 6006, orchestrator 110 can proceed to block 6008. At block 6008, orchestrator 110 can virtually reprovision and digitally adjust a configuration for the target computing environment); and
cause the at least one predicted action to be performed to modify the current state of the processing environment to a different state of the processing environment (paragraphs [0102], [6018], [1106], [1302]: sending the provisioning data/actions from the orchestrator to manager/agent components that carry out reprovisioning (deploying nodes/VMs, adjusting allocations)).
Regarding claim 23, Marzorati discloses
further comprising: wherein the one or more predicted actions comprise at least one of modifying a number of workloads being performed by the computing system or modifying hardware resources of the computing system (paragraph [0027]: Edge network disposed computing environments … can run a set of common software applications … Certain software applications … require low latency performance demanded by UE device clients being serviced by the running of the software applications; paragraph [0102]: orchestrator 110 can send provisioning data to the target computing environment of computing environments 120A-120Z and that target computing environment at provisioning block 1302 can be reprovisioned according to the provision data, e.g., by adjustment of a count of computing nodes, adjustment of computing node types, adjustment of virtualization parameter values expressed, e.g., by increases in numbers (a count) of virtual machines, adjustment of virtual machine types, and/or adjustment of virtual machine resource provisioning).
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 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 of this title, 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 2, 3, 6 are rejected under 35 U.S.C. 103 as being unpatentable over Marzorati et al. (US 2023/0236871, hereinafter Marzorati) in view of Chitalwala et al. (US 2023/0043579, hereinafter Chitalwala).
Regarding claim 2, Marzorati does not disclose wherein using the one or more machine learning models to output the one or more predicted actions further comprises obtaining one or more metrics that indicate at least one relationship between the values of the at least one operating parameter of the computing system and the values of the at least one system objective, and using the one or more metrics to select the one or more predicted actions.
Chitalwala discloses wherein using the one or more machine learning models to output the one or more predicted actions further comprises obtaining one or more metrics that indicate at least one relationship between the values of the at least one operating parameter of the computing system and the values of the at least one system objective (Fig. 4, 5; paragraphs [0030]-[0039]; an explicit performance metric OER that reflects a relationship among operating parameters (job load LT, job concurrency LC, pending jobs LW, utilization LU/R) and the objective of optimal resource utilization meeting SLA criteria; paragraph 0047): comparing first and second performance metrics and alert/difference thresholds to detect change), and using the one or more metrics to select the one or more predicted actions (paragraph [0049]-[0051]: uses the metric/threshold logic to choose concrete actions: if the metric/threshold conditions indicate over- or under-utilization relative to targets, the system triggers upsizing/downsizing of resources or adjusts workload (e.g., increase/decrease job load via load balancing). It would have been obvious to one of ordinary skill in the art at the time the claimed invention was effectively filed to incorporate Chitalwala’s metric-based action selection into Marzorati’s model-driven orchestrator so that the orchestrator obtains one or more metrics indicating relationships between operating parameters and objectives and uses those metrics to select predicted actions, because both references address automated resource management against SLA/performance goals in cloud/edge environment. The motivation would have been to monitor and optimize computing resource usage of an cloud based computing application (Chitalwala paragraph [0003]).
Regarding claim 3, Marzorati does not disclose wherein the one or more metrics comprise one or more cross-correlations between the values of the at least one operating parameter and the values of the at least one system objective.
Chitalwala discloses wherein the one or more metrics comprise one or more cross-correlations between the values of the at least one operating parameter and the values of the at least one system objective (Figs. 4, 5; paragraphs [0030]-[0039], [0048]: formulating and using a performance metric (Overall Economic Rate, OER) computed from operating parameters (job load L_T, job concurrency L_C, pending jobs L_W) and resource utilization (L_U/R) and shows correlations across time periods (September vs October); Note: Under the BRI of “cross-correlations” (metrics that quantify relationships between two classes of variables, including temporally compared correlations / difference measures), Chitalwala’s correlation plots and threshold/difference comparisons are reasonably read on the claimed “cross-correlations.”). It would have been obvious to one of ordinary skill in the art at the time the claimed invention was effectively filed to use Chitalwala’s correlation/difference metrics within Marzorati’s ML provisioning/orchestration framework to compute metrics that indicate relationships between operating parameters and system objectives and to select reprovisioning actions based on those metrics, because both references address automated resource management against SLA objectives and using correlation/threshold metrics to trigger corrective action is a routine design choice. The motivation would have been to monitor and optimize computing resource usage of an cloud based computing application (Chitalwala paragraph [0003]).
Regarding claim 6, Marzorati discloses
wherein using the one or more machine learning models to output the one or more predicted actions comprises: identifying a plurality of predicted actions using the one or more metrics, the plurality of predicted actions comprising the one or more predicted actions (paragraph [0085]: At block 6006, orchestrator 110 can query global CEs predictive model 4002 … for return of target computing environment infrastructure and/or virtualization configuration adjustment data; paragraph [0095]: On completion of block 6010, orchestrator 110 can proceed to block 6012. At block 6012, orchestrator 110 can discover simulation scenarios for testing the virtually reprovisioned target computing environment virtually reprovisioned with returned data returned at block 6006; Note: Marzorati teaches generating multiple candidate provisioning adjustments (i.e., a plurality of candidate actions) and using predictive models to evaluate them), wherein using the one or more metrics to select the one or more predicted actions comprises predicting sets of one or more potential future states using the plurality of predicted actions (paragraphs [0088]: At block 6006, orchestrator 110 on querying global CEs predictive model 4002 with the described query data … can obtain return data that specifies infrastructure parameter values and/or virtualization parameter values that are predicted to support the referenced utilization parameter values including availability parameter values input as query data at a future time period; paragraph [0097]: At block 6014, orchestrator 110 can run the identified simulation scenarios using a computing environment predictive model 3002 associated to the identified matching computing environment … Running a simulation scenario, block 6014 can include applying as query data to the identified CE predictive model 3002 … query data that comprises computing environment characterizing parameter values associated with the boundary condition; Note: Marzorati teaches applying each candidate action (candidate reprovisioning) to predictive models to obtain predicted future states (simulated outcomes) for that candidate), and selecting the one or more predicted actions for which a selected one of the sets was predicted that more closely matches the at least one desired state than at least one other of the sets (paragraph [0090]: Based on the return data returned at block 6006, orchestrator 110 can proceed to block 6008. At block 6008, orchestrator 110 can virtually reprovision and digitally adjust a configuration for the target computing environment).
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Marzorati et al. (US 2023/0236871, hereinafter Marzorati) in view of Chitalwala et al. (US 2023/0043579, hereinafter Chitalwala) as applied to claim 1, and further in view of Jain et al. (US 2024/0007414, hereinafter Jain).
Regarding claim 4, Marzorati in view of Chitalwala does not disclose wherein first
inputs comprise one or more first embeddings generated using the values of the at least one operating parameter of the computing system, and the second inputs comprise one or more second embeddings generated using the values of the at least one system objective.
Jain discloses wherein first inputs comprise one or more first embeddings generated using the values of the at least one operating parameter of the computing system (paragraphs [0220]: At least one benefit of the packaged workload ID5_A110 is that examples disclosed herein include and/or otherwise embed additional semantic information into the workload so that on-the-fly decisions can occur in view of dynamic conditions during runtime ID5_A112. Examples disclosed herein retrieve, receive and/or otherwise obtain SLA information/parameters ID5_A114 and current utilization information ID5_A116); paragraph [0223]: Knobs include, but are not limited to particular target hardware device preferences (e.g., CPU, GPU, FPGA, accelerator, particular CPU core selection(s), uncore frequencies, memory bandwidth, cache partitioning/reservation, FPGA RTL partitioning, GPU execution units partitioning/reservation, etc.), and particular optimization parameter preferences (e.g., improved latency, improved energy consumption, improved accuracy, etc.), and the second inputs comprise one or more second embeddings generated using the values of the at least one system objective (paragraph [0220]: As used herein, SLA parameters represent constraints to be satisfied by workload execution, such as accuracy metrics, speed metrics, power consumption metrics, cost (e.g., financial cost, processor burden cost, etc.) metrics, etc.; paragraph [0223]: FIG. ID5_B illustrates example graph semantic embedding for a graph of interest ID5_B200.… Different embeddings capture corresponding probabilities of certain paths in a graph towards the accuracy of the network … Based on the available information, dynamic decisions are made to schedule part of subgraph on different available compute devices in a platform. Optimized graphs ID5_B200 disclosed herein consider more than just a generic parameter in view of a predetermined computing device, but also incorporate semantic information that can trigger dynamic decisions of (a) candidate target computing devices to best handle the workload and (b) candidate graphs to best facilitate workload operation in view of dynamic SLA information, dynamic computing resource conditions, and key performance indicator (KPI) information). It would have been obvious to one of ordinary skill in the art at the time the claimed invention was effectively filed to incorporate Jain’s practice of representing operating-parameter values and system-objective values as learned embeddings (first and second inputs, respectively) into the Marzorati in view of Chitalwala system in order to enable on-the-fly, ML-driven decisions that adapt to dynamic runtime conditions and satisfy SLA targets. Jain expressly teaches embedding semantic/parameter information “so that on-the-fly decisions can occur in view of dynamic conditions during runtime” (paragraph [0220]), using “different embeddings” to drive decisions that account for “dynamic SLA information” and current resource conditions (paragraphs [0222]-[0223]), and retrieving/acting on current SLA/utilization at runtime to select paths that “exhibit predicted SLA compliance” (paragraph [0228-[0229]). The motivation would have been to improve execution performance across the multiple AI models in a multi-tenant computing system (Jain paragraph [0276]).
Claims 16 and 28 are rejected under 35 U.S.C. 103 as being unpatentable over Marzorati et al. (US 2023/0236871, hereinafter Marzorati) in view of Jain et al. (US 2024/0007414, hereinafter Jain).
Regarding claims 16 and 28, Marzorati does not disclose wherein the first inputs comprise one or more first embeddings generated using the values of the one or more parameters, and the second inputs comprise one or more second embeddings generated using the values of the one or more objectives.
Jain discloses wherein the first inputs comprise one or more first embeddings generated using the values of the one or more parameters (paragraphs [0220]: At least one benefit of the packaged workload ID5_A110 is that examples disclosed herein include and/or otherwise embed additional semantic information into the workload so that on-the-fly decisions can occur in view of dynamic conditions during runtime ID5_A112. Examples disclosed herein retrieve, receive and/or otherwise obtain SLA information/parameters ID5_A114 and current utilization information ID5_A116); paragraph [0223]: Knobs include, but are not limited to particular target hardware device preferences (e.g., CPU, GPU, FPGA, accelerator, particular CPU core selection(s), uncore frequencies, memory bandwidth, cache partitioning/reservation, FPGA RTL partitioning, GPU execution units partitioning/reservation, etc.), and particular optimization parameter preferences (e.g., improved latency, improved energy consumption, improved accuracy, etc.), and the second inputs comprise one or more second embeddings generated using the values of the one or more objectives (paragraph [0220]: As used herein, SLA parameters represent constraints to be satisfied by workload execution, such as accuracy metrics, speed metrics, power consumption metrics, cost (e.g., financial cost, processor burden cost, etc.) metrics, etc.; paragraph [0223]: FIG. ID5_B illustrates example graph semantic embedding for a graph of interest ID5_B200.… Different embeddings capture corresponding probabilities of certain paths in a graph towards the accuracy of the network … Based on the available information, dynamic decisions are made to schedule part of subgraph on different available compute devices in a platform. Optimized graphs ID5_B200 disclosed herein consider more than just a generic parameter in view of a predetermined computing device, but also incorporate semantic information that can trigger dynamic decisions of (a) candidate target computing devices to best handle the workload and (b) candidate graphs to best facilitate workload operation in view of dynamic SLA information, dynamic computing resource conditions, and key performance indicator (KPI) information). It would have been obvious to one of ordinary skill in the art at the time the claimed invention was effectively filed to incorporate Jain’s practice of representing operating-parameter values and system-objective values as learned embeddings (first and second inputs, respectively) into the Marzorati in view of Chitalwala system in order to enable on-the-fly, ML-driven decisions that adapt to dynamic runtime conditions and satisfy SLA targets. Jain expressly teaches embedding semantic/parameter information “so that on-the-fly decisions can occur in view of dynamic conditions during runtime” (paragraph [0220]), using “different embeddings” to drive decisions that account for “dynamic SLA information” and current resource conditions (paragraphs [0222]-[0223]), and retrieving/acting on current SLA/utilization at runtime to select paths that “exhibit predicted SLA compliance” (paragraph [0228-[0229]). The motivation would have been to improve execution performance across the multiple AI models in a multi-tenant computing system (Jain paragraph [0276]).
Claim 14 is rejected under 35 U.S.C. 103 as being unpatentable over Marzorati et al. (US 2023/0236871, hereinafter Marzorati) in view of Raghunath et al. (US 2022/0116455, hereinafter Raghunath).
Regarding claim 14,
Marzorati does not disclose wherein the one or more circuits comprise a data processing unit (“DPU”). Raghunath discloses wherein the one or more circuits comprise a data processing unit (“DPU”) (paragraph [0067]: the processor 604 may be embodied as a specialized x-processing unit (x-PU) also known as a data processing unit (DPU) … ). It would have been obvious to one of ordinary skill in the art at the time the claimed invention was effectively filed to incorporate Raghunath’s DPU into Marzorati’s processor. The motivation would have been to process one or more data streams and perform specific tasks and actions for the data streams (such as hosting microservices, performing service management or orchestration, organizing or managing server or data center hardware, managing service meshes, or collecting and distributing telemetry), outside of the CPU or general purpose processing hardware (Raghunath paragraph [0067]).
Allowable Subject Matter
Claims 5, 15, 17-20, 24-27 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in [0037] 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 extension fee pursuant to [0037] 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 date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SISLEY N. KIM whose telephone number is (571)270-7832. The examiner can normally be reached M-F 11:30AM -7:30PM.
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, April Y. Blair can be reached on (571)270-1014. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300. 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.
/SISLEY N KIM/Primary Examiner, Art Unit 2196 4/18/2026
Prosecution Insights