DETAILED ACTION
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 Amendment
Applicant's amendment filed on 01/05/2026 has been entered. Claims 1, 4, 7, 9-15, 17-18 and 20 have been amended. No claims have been added or cancelled. Claims 1-20 are still pending in this application, with claims 1, 9 and 15, being independent.
Response to Arguments
Applicant's arguments filed 01/05/2026 have been fully considered but they are not persuasive.
Regarding independent claim 1, Applicant submits that the combination of Schmidt and Guttman does not specifically teach the newly-added claim limitations.
However, the Examiner respectfully disagrees with the Applicant and asserts that the combination of Schmidt and Guttman still reads on the newly-added claim limitations.
Specifically, Schmidt, in Par. 0071, teaches that the dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering). In addition, the user also specifies to the dispatcher which variables are input and which are output variables in the dataset. Stopping criterion is a user specified criterion telling HAMLET when to stop AutoML. This can be in the form of a time budget, a minimum required performance threshold (related to the loss function) for the user's application, or lack of improvement (e.g. a period of time or number of main loop iterations) during which HAMLET's tuners did not improve in performance anymore. Based at least from this section, Schmidt teaches the claim limitation “…determining, by the terminal device, a value of a performance measurement index based on the first data, wherein the performance measurement index indicates validity of the first selection method for the ML task…” as argued by the Applicant.
Further, in Figure 12 and Par. 0198 and 203 of Guttman, “providing the estimated processing resources requirements and/or information based on the estimated processing resources requirements to a user, to another process, to an external device, and so forth. In some examples, the information may be transmitted to an external device, for example using communication devices” still reads on the newly-added claim limitation “sending, by the terminal device, the value of the performance measurement index to the network device”.
Given, the current rejection and having the additional reasoning as presented above, the combination of Schmidt and Guttman reads upon the claimed invention of claim 1.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim(s) 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Schmidt et al. (US 2021/0224585; hereinafter Schmidt) in view of Guttman (US 2018/0336479; hereinafter Guttman).
Regarding claim 1, Schmidt shows a data transmission method (Figure 1 shows the system performing the disclosed method of selecting a machine learning algorithm and transmission of such selection.), comprising:
receiving, by a device, information about a first selection method that indicates a method for selecting first data associated with a machine learning (ML) task (Figure 1; Par. 0056-0059, 0071, 0075-0076; receiving, through the dispatcher, upload a data set (machine learning task per Par. 0071) and define a data set description. The dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering). The user also specifies to the dispatcher which variables are input and which are output variables in the dataset.);
performing, by the device, data selection of data stored in the terminal device by using the first selection method, to obtain the first data for the ML task (Figure 1; Par. 0073-0074, 0078, 0078; when presented a new user task, HAMLET can choose to disable arms that do not fit the task or recommend arms which are especially promising. HAMLET uses features describing the task's dataset, e.g., the size, the data types present in the dataset, if there are missing features, etc. Also, the type of machine learning task (e.g. regression vs clustering) is meaningful to consider for the decision about which arms to apply. The dataset is stored in a dataset storage.);
determining, by the terminal device, a value of a performance measurement index based on the first data, wherein the performance measurement index indicates validity of the first selection method for the ML task (Par. 0071, 0098-0099; Table 1; dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering). The user also specifies to the dispatcher which variables are input and which are output variables in the dataset. Stopping criterion is a user specified criterion telling HAMLET when to stop AutoML. This can be in the form of a time budget, a minimum required performance threshold (related to the loss function) for the user's application, or lack of improvement (e.g. a period of time or number of main loop iterations) during which HAMLET's tuners did not improve in performance anymore.);
sending, by the device, the first data (Figure 1; Par. 0056, 0060-0063, 0095-0096; dispatcher transmits information on the progress of a specified machine learning task, e.g., performance scores and elapsed budget, optionally transmits a notification when the training finished, transmits an indication which was/were the best performing model(s) or ensemble(s); optionally transmits the trained models as, e.g., serialized binary objects in a standardized format such as pickled python objects, and transmits references to the models within HAMLET such as unique identifiers of the models within a HAMLET database.).
Schmidt shows all of the elements as discussed above. Schmidt does not specifically show receiving by a terminal device from a network device, information associated with a machine learning (ML) task, sending the first data to the network device and sending, by the terminal device, the value of the performance measurement index to the network device.
However, the above-mentioned claim limitations are well-established in the art as evidenced by Guttmann. Specifically, Guttmann shows receiving by a terminal device from a network device, information associated with a machine learning (ML) task (Figure 12; Par. 0198, 0203; obtaining properties of a machine learning training task (Step 1210) may comprise obtaining one or more records describing one or more machine learning training tasks. At least part of the properties of a machine learning training task may be received from external devices, for example using communication devices.), sending the first data to the network device and sending, by the terminal device, the value of the performance measurement index to the network device (Figure 12; Par. 0198, 0203; Providing the estimation (Step 1240) may comprise providing the estimated processing resources requirements of Step 1230 and/or information based on the estimated processing resources requirements to a user, to another process, to an external device, and so forth. In some examples, the information may be transmitted to an external device, for example using communication devices.).
In view of the above, having the system of Schmidt, then given the well-established teaching of Guttmann, it would have been obvious before the effective filing date of the claimed invention to modify the system of Schmidt as taught by Guttmann, in order to provide motivation for replacing the usage of one dataset with another dataset, replacing the usage of a first set of annotations with a second set, and so forth, for example with smaller datasets and/or annotations that are predicted to shorten the processing time (Par. 0284 of Guttmann).
Regarding claim 2, modified Schmidt shows wherein the method further comprises: sending, by the terminal device to the network device, capability information for determining the first selection method (Schmidt: Par. 0056, 0062-0064; dispatcher transmits information on the progress of a specified machine learning task, e.g., performance scores and elapsed budget, optionally transmits a notification when the training finished, transmits an indication which was/were the best performing model(s) or ensemble(s); optionally transmits the trained models as, e.g., serialized binary objects in a standardized format such as pickled python objects, and transmits references to the models within HAMLET such as unique identifiers of the models within a HAMLET database.).
Regarding claim 3, modified Schmidt shows wherein the capability information comprises at least one of the following:
a computing capability comprising a data processing capability of the terminal device;
a data capacity comprising an amount of data stored by the terminal device for ML; and
a second selection method comprising a selection method supported by the terminal device (Schmidt: Par. 0056, 0062-0064; dispatcher transmits information on the progress of a specified machine learning task, e.g., performance scores and elapsed budget, and optionally transmits the trained models as, e.g., serialized binary objects in a standardized format such as pickled python objects.).
Regarding claim 4, modified Schmidt shows before determining the value of the performance index, receiving, by the terminal device, a performance measurement index from the network device (Schmidt: Par. (Par. 0071, 0098-0099; Table 1; dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering).).
Regarding claim 5, modified Schmidt shows wherein the performance measurement index comprises at least one of the following: a Jaccard coefficient (JC), a Rand index (RI), an accuracy rate (Schmidt: Table – Algorithm 1.2; noted user define accuracy.), or an error rate.
Regarding claim 6, modified Schmidt shows wherein the method further comprises: receiving, by the terminal device from the network device, a stopping criterion indicating a condition that needs to be met when the terminal device stops the data selection; and the performing, by the terminal device, data selection by using the first selection method, to obtain first data comprises: performing, by the terminal device, the data selection based on the stopping criterion (Schmidt: Table 1 includes a stopping criterion defined as a user specified criterion telling HAMLET when to stop AutoML. This can be in the form of a time budget, a minimum required performance threshold (related to the loss function) for the user's application, or lack of improvement (e.g. a period of time or number of main loop iterations) during which HAMLET's tuners did not improve in performance anymore. The HAMLET task includes the dataset to be learned upon, the type of learning problem regression, classification or clustering and the loss function can be specified, also a stopping criterion can be specified.).
Regarding claim 7, modified Schmidt shows wherein the stopping criterion comprises at least one of the following:
a feature quantity comprising a quantity of features of the first data;
a quantity of iterations comprising a quantity of iterations performed by the terminal device by using the first selection method during the data selection (Schmidt: Table 1 includes a stopping criterion defined as a user specified criterion telling HAMLET when to stop AutoML. This can be in the form of a time budget, a minimum required performance threshold (related to the loss function) for the user's application, or lack of improvement (e.g. a period of time or number of main loop iterations) during which HAMLET's tuners did not improve in performance anymore. The HAMLET task includes the dataset to be learned upon, the type of learning problem regression, classification or clustering and the loss function can be specified, also a stopping criterion can be specified.); and
the iteration performance condition comprising a value of the performance measurement index that needs to be satisfied when the terminal device stops the data selection by using the first selection method (Schmidt: Table 1 includes a stopping criterion defined as a user specified criterion telling HAMLET when to stop AutoML. This can be in the form of a time budget, a minimum required performance threshold (related to the loss function) for the user's application, or lack of improvement (e.g. a period of time or number of main loop iterations) during which HAMLET's tuners did not improve in performance anymore. The HAMLET task includes the dataset to be learned upon, the type of learning problem regression, classification or clustering and the loss function can be specified, also a stopping criterion can be specified.).
Regarding claim 8, modified Schmidt shows wherein the first selection method comprises one of the following: a relevant features method (Schmidt: Par. 0065; The dispatcher registers tasks with the MasterBandit and invokes it to start training. For that, the dispatcher passes relevant configuration parameters (e.g., dataset reference task description, arm configurations) via interface B.), an information gain method, a multi-cluster feature selection method (Schmidt: Par. 0071; The dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering).), a Las Vegas wrapper method, a ridge regression method (Schmidt: Par. 0071; The dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering).), or a least absolute shrinkage operator method.
Regarding claim 9, Schmidt shows a communication apparatus (Figure 1 shows the system performing the disclosed method of selecting a machine learning algorithm and transmission of such selection. System implemented as a tangible, non-transitory computer-readable medium having instructions thereon which, upon being executed by one or more processors, alone or in combination, provide for execution of the disclosed method as shown in Par. 0047.), comprising:
at least one processor; one or more memories coupled to the at least one processor and storing programming instructions to perform operations (System implemented as a tangible, non-transitory computer-readable medium having instructions thereon which, upon being executed by one or more processors, alone or in combination, provide for execution of the disclosed method as shown in Par. 0047.) comprising:
receiving information about a first selection method that indicates a method for selecting first data associated with a machine learning (ML) task (Figure 1; Par. 0056-0059, 0071, 0075-0076; receiving, through the dispatcher, upload a data set (machine learning task per Par. 0071) and define a data set description. The dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering). The user also specifies to the dispatcher which variables are input and which are output variables in the dataset.);
performing, by the device, data selection of data stored in the terminal device by using the first selection method, to obtain the first data for the ML task (Figure 1; Par. 0073-0074, 0078, 0078; when presented a new user task, HAMLET can choose to disable arms that do not fit the task or recommend arms which are especially promising. HAMLET uses features describing the task's dataset, e.g., the size, the data types present in the dataset, if there are missing features, etc. Also, the type of machine learning task (e.g. regression vs clustering) is meaningful to consider for the decision about which arms to apply. The dataset is stored in a dataset storage.);
determining, by the terminal device, a value of a performance measurement index based on the first data, wherein the performance measurement index indicates validity of the first selection method for the ML task (Par. 0071, 0098-0099; Table 1; dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering). The user also specifies to the dispatcher which variables are input and which are output variables in the dataset. Stopping criterion is a user specified criterion telling HAMLET when to stop AutoML. This can be in the form of a time budget, a minimum required performance threshold (related to the loss function) for the user's application, or lack of improvement (e.g. a period of time or number of main loop iterations) during which HAMLET's tuners did not improve in performance anymore.); and
sending the first data (Figure 1; Par. 0056, 0060-0063, 0095-0096; dispatcher transmits information on the progress of a specified machine learning task, e.g., performance scores and elapsed budget, optionally transmits a notification when the training finished, transmits an indication which was/were the best performing model(s) or ensemble(s); optionally transmits the trained models as, e.g., serialized binary objects in a standardized format such as pickled python objects, and transmits references to the models within HAMLET such as unique identifiers of the models within a HAMLET database.).
Schmidt shows all of the elements as discussed above. Schmidt does not specifically show receiving by a terminal device from a network device, information associated with a machine learning (ML) task, sending the first data to the network device and sending, by the terminal device, the value of the performance measurement index to the network device.
However, the above-mentioned claim limitations are well-established in the art as evidenced by Guttmann. Specifically, Guttmann shows receiving by a terminal device from a network device, information associated with a machine learning (ML) task (Figure 12; Par. 0198, 0203; obtaining properties of a machine learning training task (Step 1210) may comprise obtaining one or more records describing one or more machine learning training tasks. At least part of the properties of a machine learning training task may be received from external devices, for example using communication devices.), sending the first data to the network device and sending, by the terminal device, the value of the performance measurement index to the network device (Figure 12; Par. 0198, 0203; Providing the estimation (Step 1240) may comprise providing the estimated processing resources requirements of Step 1230 and/or information based on the estimated processing resources requirements to a user, to another process, to an external device, and so forth. In some examples, the information may be transmitted to an external device, for example using communication devices.).
In view of the above, having the system of Schmidt, then given the well-established teaching of Guttmann, it would have been obvious before the effective filing date of the claimed invention to modify the system of Schmidt as taught by Guttmann, in order to provide motivation for replacing the usage of one dataset with another dataset, replacing the usage of a first set of annotations with a second set, and so forth, for example with smaller datasets and/or annotations that are predicted to shorten the processing time (Par. 0284 of Guttmann).
Regarding claims 10, 11, 12, 13 and 14, these claims are rejected based on the same reasoning as presented in the rejection of claims 2, 3, 4, 5 and 6, respectively.
Regarding claim 15, Schmidt shows a non-transitory computer-readable storage medium storing one or more instructions executable by at least one processor to perform operations (Figure 1 shows the system performing the disclosed method of selecting a machine learning algorithm and transmission of such selection. System implemented as a tangible, non-transitory computer-readable medium having instructions thereon which, upon being executed by one or more processors, alone or in combination, provide for execution of the disclosed method as shown in Par. 0047.) comprising:
receiving information about a first selection method that indicates a method for selecting first data associated with a machine learning (ML) task (Figure 1; Par. 0056-0059, 0071, 0075-0076; receiving, through the dispatcher, upload a data set (machine learning task per Par. 0071) and define a data set description. The dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering). The user also specifies to the dispatcher which variables are input and which are output variables in the dataset.);
performing, by the device, data selection of data stored in the terminal device by using the first selection method, to obtain the first data for the ML task (Figure 1; Par. 0073-0074, 0078, 0078; when presented a new user task, HAMLET can choose to disable arms that do not fit the task or recommend arms which are especially promising. HAMLET uses features describing the task's dataset, e.g., the size, the data types present in the dataset, if there are missing features, etc. Also, the type of machine learning task (e.g. regression vs clustering) is meaningful to consider for the decision about which arms to apply. The dataset is stored in a dataset storage.);
determining a value of a performance measurement index based on the first data, wherein the performance measurement index indicates validity of the first selection method for the ML task (Par. 0071, 0098-0099; Table 1; dispatcher receives a machine learning task referring a dataset, a description of the dataset, a loss function, a stopping criterion and the type of machine learning problem the task is about (regression, classification or clustering). The user also specifies to the dispatcher which variables are input and which are output variables in the dataset. Stopping criterion is a user specified criterion telling HAMLET when to stop AutoML. This can be in the form of a time budget, a minimum required performance threshold (related to the loss function) for the user's application, or lack of improvement (e.g. a period of time or number of main loop iterations) during which HAMLET's tuners did not improve in performance anymore.); and
sending the first data (Figure 1; Par. 0056, 0060-0063, 0095-0096; dispatcher transmits information on the progress of a specified machine learning task, e.g., performance scores and elapsed budget, optionally transmits a notification when the training finished, transmits an indication which was/were the best performing model(s) or ensemble(s); optionally transmits the trained models as, e.g., serialized binary objects in a standardized format such as pickled python objects, and transmits references to the models within HAMLET such as unique identifiers of the models within a HAMLET database.).
Schmidt shows all of the elements as discussed above. Schmidt does not specifically show receiving by a terminal device from a network device, information associated with a machine learning (ML) task and transmitting first data to the network device.
However, the above-mentioned claim limitations are well-established in the art as evidenced by Guttmann. Specifically, Guttmann shows receiving by a terminal device from a network device, information associated with a machine learning (ML) task and transmitting first data to the network device (Figure 12; Par. 0198, 0203; obtaining properties of a machine learning training task (Step 1210) may comprise obtaining one or more records describing one or more machine learning training tasks. At least part of the properties of a machine learning training task may be received from external devices, for example using communication devices. Providing the estimation (Step 1240) may comprise providing the estimated processing resources requirements of Step 1230 and/or information based on the estimated processing resources requirements to a user, to another process, to an external device, and so forth. In some examples, the information may be transmitted to an external device, for example using communication devices.).
In view of the above, having the system of Schmidt, then given the well-established teaching of Guttmann, it would have been obvious before the effective filing date of the claimed invention to modify the system of Schmidt as taught by Guttmann, in order to provide motivation for replacing the usage of one dataset with another dataset, replacing the usage of a first set of annotations with a second set, and so forth, for example with smaller datasets and/or annotations that are predicted to shorten the processing time (Par. 0284 of Guttmann).
Regarding claims 16, 17, 18 and 19, these claims are rejected based on the same reasoning as presented in the rejection of claims 2, 3, 4 and 5, respectively.
Regarding claim 20, this claim is rejected based on the same reasoning as presented in the rejection of claim 6.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 11227047 B1 - The presently disclosed subject matter includes an apparatus that receives a dataset with values associated with different digital resources captured from a group of compute devices.
US 20200257961 A1 - A method for determining an architecture for a task neural network configured to perform a particular machine learning task is described.
US 20170124486 A1 - In accordance with disclosed embodiments, there are provided systems, methods, and apparatuses for implementing predictive engine evaluation and replay of engine performance.
US 20240005129 A1 - Methods, systems, and apparatus, including computer programs encoded on computer storage media, for jointly determining neural network architectures and hardware accelerator architectures.
US 20200012584 A1 - in an embodiment, a method for optimizing a model may comprise receiving a data input that includes a desired outcome and an input dataset identifier.
US 20240112090 A1 - Certain aspects of the present disclosure provide techniques for concurrently performing inferences using a machine learning model and optimizing parameters used in executing the machine learning model.
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 REDENTOR M PASIA whose telephone number is (571)272-9745. The examiner can normally be reached Mondays-Thursdays - 5am-245pm and Fridays 5am-330pm.
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, Un Cho can be reached at (571)272-7919. 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.
/REDENTOR PASIA/Primary Examiner, Art Unit 2413
Prosecution Insights