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 .
Claims 1-20 are presented for examination.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on February 7th, 2025, is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Specification
The disclosure is objected to because of the following informalities:
Paragraphs 4 and 20: "is become more prevalent" should read "is becoming more prevalent".
Paragraph 17: "operations that are performed are caused to be performed" should read "operations that are performed or are caused to be performed"
Paragraphs 21, 41, and 44: "publically available" should read "publicly available".
Paragraph 43: "and new data labeling module may provide the decrypted new training dataset directly to ML model" should read "and the new data labeling module may provide the decrypted new training dataset directly to the ML model"
Paragraph 46: "and new data labeling module 108 may provide the decrypted new training dataset directly to ML model 100" should read "and the new data labeling module 108 may provide the decrypted new training dataset directly to the ML model 100"
Paragraph 50: "the new datasets used by the end users is usually confidential" should read "the new datasets used by the end users are usually confidential"
Paragraph 50: "recommend it to the ends users" should read "recommend it to the end users"
Appropriate correction is required.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitation(s) is/are:
"a data label module" in claims 1, 8, and 15.
"a new data labeling module" in claims 2, 9, and 15.
"a well evaluation module" in claims 6, 13, and 19.
Regarding the invocation of 112f, see rejections under 35 U.S.C. 112(a) and 112(b) infra.
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
The claim limitation "data label module" of claims 1, 8, and 15 invokes 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. However, the written description fails to disclose the corresponding structure, material, or acts for performing the entire claimed function and to clearly link the structure, material, or acts to the function. At most, the specification at paragraphs 42, 45, and 49 indicates that the data label module (e.g., data label module 60B) is "provided to end users as a software update to enable the end users to label their own datasets," which does not disclose any hardware structure or algorithm to perform the claimed labeling function.
The claim limitation "new data labeling module" of claims 2, 9, and 15 invokes 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. However, the written description fails to disclose the corresponding structure, material, or acts for performing the entire claimed function and to clearly link the structure, material, or acts to the function. At most, the specification at paragraphs 43 and 46 indicates that the new data labeling module (e.g., new data labeling module 108) "may include logic to decrypt the new training dataset, for example, using an encryption key specific to an end user," without providing any algorithm (e.g., mathematical formula, sequence of steps) for performing the claimed decrypting function.
The claim limitation "well evaluation module" of claims 6, 13, and 19 invokes 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. However, the written description fails to disclose the corresponding structure, material, or acts for performing the entire claimed function and to clearly link the structure, material, or acts to the function. At most, the specification at paragraph 44 describes a well integrity evaluation module 60A that is launched "to present data relating to the prediction results," without disclosing any hardware structure or algorithm to perform the claimed presenting function.
Thus, the specification does not disclose a definite structure for the "data label module," the "new data labeling module," and the “well evaluation module,” and does not provide an accompanying algorithm to perform the claimed specific computer function. It has been recognized by the courts that "merely restating a function associated with a means-plus-function limitation is insufficient to provide the corresponding structure for definiteness" (see MPEP 2181 (IV)). Therefore, it is unclear whether Applicant had possession of the claimed invention as of the effective filing date. For the purposes of further examination, the "data label module," the "new data labeling module," and the “well evaluation module” will be interpreted as part of a computer processor.
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
The claim limitation "data label module" of claims 1, 8, and 15 invokes 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. However, the written description fails to disclose the corresponding structure, material, or acts for performing the entire claimed function and to clearly link the structure, material, or acts to the function. At most, the specification at paragraphs 42, 45, and 49 indicates that the data label module (e.g., data label module 60B) is "provided to end users as a software update to enable the end users to label their own datasets," which does not disclose any hardware structure or algorithm to perform the claimed labeling function.
The claim limitation "new data labeling module" of claims 2, 9, and 15 invokes 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. However, the written description fails to disclose the corresponding structure, material, or acts for performing the entire claimed function and to clearly link the structure, material, or acts to the function. At most, the specification at paragraphs 43 and 46 indicates that the new data labeling module (e.g., new data labeling module 108) "may include logic to decrypt the new training dataset, for example, using an encryption key specific to an end user," without providing any algorithm (e.g., mathematical formula, sequence of steps) for performing the claimed decrypting function.
The claim limitation "well evaluation module" of claims 6, 13, and 19 invokes 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. However, the written description fails to disclose the corresponding structure, material, or acts for performing the entire claimed function and to clearly link the structure, material, or acts to the function. At most, the specification at paragraph 44 describes a well integrity evaluation module 60A that is launched "to present data relating to the prediction results," without disclosing any hardware structure or algorithm to perform the claimed presenting function.
Thus, the specification does not disclose a definite structure for the "data label module," the "new data labeling module," and the "well evaluation module," and does not provide an accompanying algorithm to perform the claimed specific computer functions. The scope of these functional limitations, therefore, encompasses any and all software that performs the claimed functions, without limitation to a specific disclosed structure or algorithm. It has been recognized by the courts that "merely restating a function associated with a means-plus-function limitation is insufficient to provide the corresponding structure for definiteness" (see MPEP 2181 (IV)). Therefore, the claims are indefinite and are rejected under 35 U.S.C. 112(b) or pre-AIA 35 U.S.C. 112, second paragraph.
In response to this rejection, applicant must clarify whether this limitation should be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. Mere assertion regarding applicant’s intent to invoke or not invoke 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph is insufficient. Applicant may:
(a) Amend the claim to clearly invoke 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, by reciting “means” or a generic placeholder for means, or by reciting “step.” The “means,” generic placeholder, or “step” must be modified by functional language, and must not be modified by sufficient structure, material, or acts for performing the claimed function;
(b) Present a sufficient showing that 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, should apply because the claim limitation recites a function to be performed and does not recite sufficient structure, material, or acts to perform that function;
(c) Amend the claim to clearly avoid invoking 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, by deleting the function or by reciting sufficient structure, material or acts to perform the recited function; or
(d) Present a sufficient showing that 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, does not apply because the limitation does not recite a function or does recite a function along with sufficient structure, material or acts to perform that function.
Claim 15 recites "encrypting, via a data label module, a training dataset," and later recites "receiving, via the cloud-based application, a training dataset specific to a well." Both limitations introduce "a training dataset," and it is unclear whether the encrypted training dataset and the received training dataset specific to a well are the same or different training datasets.
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.
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.
Claims 1, 3, 6, 8, 10, and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Saghir (US 20220316314 A1) in view of Asher (US 11663517 B2).
Regarding claim 1,
Saghir teaches training a machine learning (ML) model to create a trained ML model, wherein the trained ML model is configured to output well prediction results based on datasets specific to respective wells (Paragraph 68 of Saghir, "the one or more ML models 224 should be appropriately trained before the models can perform their functions effectively. Training may be done using a self-supervised learning method in which labeled historical data is applied as an input to the ML models.", Paragraph 59, "embodiments of the present disclosure provide methods and systems for developing and deploying robust ML modeling in an IIoT environment to automatically predict failures and non-optimal operations directly at the well site", Paragraph 110, "the data used to build this confusion matrix 2400 was obtained from 4 wells over a period of approximately 1 month."
Saghir trains one or more ML models on labeled well-site data to predict failures and non-optimal operations at the well site. The models are trained and applied on data obtained from respective wells, which corresponds to well prediction results based on datasets specific to respective wells.)
providing a data label module to a computing system associated with an end user (Paragraph 66 of Saghir, "A well site monitoring and control application 220, or rather the computer-readable instructions therefor, may also reside on or be downloaded to the storage device 210. The well site monitoring and control application 220… to perform ML-based analytics on dynagraphs from the RTU 130 to automatically detect abnormal operations and generate a response thereto.”, Paragraph 83, "For example, the ML models may be unable to identify a dynagraph card that was not included in the training phase if the card is unique to a specific pump behavior. As another example, the ML models may be unable to make a clear distinction between two dynagraph card types that are similar to each other, such as “gas interference” and “gas lock” dynagraph cards. In such cases, the dynagraph cards may be labeled locally by a field operator via a locally available HMI.", Paragraph 84, "In this example, the user interface 1000 includes several navigation buttons, indicated generally at 1002, that allow an operator to navigate and otherwise scroll through and access various functions available on the edge gateway. A dynagraph area 1004 displays the dynagraph of interest, which may be the current dynagraph or an older dynagraph…the operator may then label the dynagraph with his own classification, which is then stored in a feedback database with the dynagraph of interest for subsequent consideration and use in model retraining and readjustment."
Saghir provides the field operator's edge computing system with a monitoring and control application having a labeling interface (HMI/GUI 1000) by which the operator labels the data. This downloaded labeling application corresponds to the claimed data label module provided to a computing system associated with an end user.)
deploying the trained ML model to a cloud-based application (Paragraph 58, “ML-based analytics is implemented at well sites in conjunction with the broader movement toward a so-called Industrial Internet of Things (IIoT). In an IIoT landscape, well site operators increasingly make use of interconnected sensors and devices to collect and store large amounts of data directly at the well sites. This data can then be processed at the well sites by an edge device equipped with ML modeling to monitor and control well operations…The edge device can then generate alarms, carry out contingent actions, and store the results of the processing to the Cloud as needed. The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations.”, Paragraph 62 of Saghir, "the data may be forwarded to other systems within an enterprise 138 and/or to the Cloud 140 (which may include a private enterprise Cloud) for further processing as needed.", Paragraph 87, “In some embodiments, the outputs from the ML models are optionally provided to an Ensemble model at 1114 to produce more accurate results. In either case, the resulting conclusions of the modeling (e.g., dynagraph classifications) are outputted at 1116, typically in terms of probabilities. This step can entail displaying the conclusions on an HMI of the edge device, storing them in a local or remote storage device or database, sending them to an external system (e.g., a SCADA host), and/or uploading them to a Cloud-computing platform for further processing.”
Saghir deploys and manages the ML models via a Cloud platform that performs further processing, which corresponds to deploying the trained ML model to a cloud-based application.)
receiving, via the cloud-based application, a training dataset specific to a well, wherein the training dataset comprises data labeling generated by the end user via the data label module (Paragraph 58, “The edge device can then generate alarms, carry out contingent actions, and store the results of the processing to the Cloud as needed. The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations.”, Paragraph 83 of Saghir, "the dynagraph cards may be labeled locally by a field operator via a locally available HMI. These manually labeled cards can then be used to retrain the models (or parts thereof)", Paragraph 84, "the operator may then label the dynagraph with his own classification, which is then stored in a feedback database with the dynagraph of interest for subsequent consideration and use in model retraining and readjustment.", Paragraph 118, "Alternatively, an edge gateway with ML-based analytics can take advantage of full Cloud connectivity where multiple such gateways can be maintained and managed via a Cloud platform. This option allows real-time data collected from multiple edge gateways to be used to retrain ML models"
The operator-labeled well-site cards are collected via the Cloud platform and used for retraining. The manually labeled dataset generated by the operator corresponds to a training dataset comprising data labeling generated by the end user via the data label module.)
re-training, via the cloud-based application, the trained ML model using the training dataset to generate a … re-trained ML model [candidate] (Paragraph 83 of Saghir, "These manually labeled cards can then be used to retrain the models (or parts thereof)", Paragraph 118, "Alternatively, an edge gateway with ML-based analytics can take advantage of full Cloud connectivity where multiple such gateways can be maintained and managed via a Cloud platform. This option allows real-time data collected from multiple edge gateways to be used to retrain ML models, resulting in more accurate applications at the edge. This retraining process can be automated and retrained models can be deployed to multiple gateways based on a preset schedule, whenever model accuracy reaches a predefined level, or when a unique dynagraph card is added (e.g., via Transfer Learning)."
Saghir re-trains the ML models on the Cloud using the newly labeled dataset in an automated retraining process. Saghir teaches the re-training via the cloud-based application using the training dataset to generate a single retrained ML model.)
outputting, from the cloud-based application, well prediction results and the … re-trained ML model to the computing system associated with the end user (Paragraph 80, “FIG. 7 shows an exemplary graphical presentation 700 of the outputs of the Ensemble model (or models) as displayed, for example, on an HMI (e.g., HMI 212) or other display of an edge gateway equipped with the ML-based analytics capability disclosed herein. The presentation 700 shows dynagraph classifications reflecting possible operating conditions for a given rod pump assembly over the course of one day based on the dynagraphs from that rod pump assembly, with the vertical axis indicating probability and the horizontal axis showing time.”, Paragraph 87 of Saghir, "In some embodiments, the outputs from the ML models are optionally provided to an Ensemble model at 1114 to produce more accurate results. In either case, the resulting conclusions of the modeling (e.g., dynagraph classifications) are outputted at 1116... This step can entail displaying the conclusions on an HMI of the edge device... sending them to an external system (e.g., a SCADA host), and/or uploading them to a Cloud-computing platform for further processing.", Paragraph 89, “the dynagraph classifications that are outputted at 1116 may be further processed and used to anticipate future pump failures. Thus, in addition to outputting the dynagraph classifications to a local or remote storage device or database, sending them to an external system, and so forth, the classifications and underlying dynagraphs may be further outputted or otherwise provided to one or more additional ML models on the edge device to anticipate future failures.”, Paragraph 118, "This retraining process can be automated and retrained models can be deployed to multiple gateways based on a preset schedule, whenever model accuracy reaches a predefined level, or when a unique dynagraph card is added (e.g., via Transfer Learning)."
Saghir outputs the prediction conclusions to the operator's edge device and deploys the retrained models back to the gateways, which corresponds to outputting the well prediction results and the re-trained ML model to the computing system associated with the end user.)
Saghir does not teach re-training, via the cloud-based application, the trained ML model using the training dataset to generate a plurality of re-trained ML model candidates or automatically evaluating the plurality of re-trained ML model candidates to generate a recommended re-trained ML model.
Asher, in the same field of endeavor, teaches [training]… the … ML model using the training dataset to generate a plurality of [trained] ML model candidates (Col. 8 Lines 38-43 of Asher, "the database server 210 may generate the predictive learning model based on training a plurality of candidate machine learning models. The plurality of candidate machine learning models may include any number of machine learning algorithms used in predictive model building.", Col. 10 Lines 28-33, “process flow 400 may also include receiving a selection of a plurality of fields 430, removing a subset of data 435, evaluation 440, removing one or more features 445, training learning models 450, evaluating learning models 455, selecting learning models 460, determining scores 465, and saving the determined scores 470.”, Col. 26 Claim 1, "generating a plurality of candidate predictive machine learning models at the server based at least in part on the plurality of features."
Asher trains, from the received data set, a plurality of candidate predictive machine learning models. Each such candidate model, when applied in Saghir's cloud-based re-training loop using the operator-labeled training dataset, corresponds to the claimed plurality of re-trained ML model candidates.)
automatically evaluating… the plurality of [trained] ML model candidates to generate a recommended [trained] ML model (Col. 8 Lines 43-50 of Asher, "The database server 210 may then evaluate the plurality of candidate machine learning models. This evaluation may be based in part on a predictive accuracy of each of the machines. The database server 210 may subsequently select the predictive machine learning model based in part on the evaluation (e.g., based on which model is most accurate, or is otherwise best suited for the selected data set or the desired predictive value).", Col. 10 Lines 28-33, “process flow 400 may also include receiving a selection of a plurality of fields 430, removing a subset of data 435, evaluation 440, removing one or more features 445, training learning models 450, evaluating learning models 455, selecting learning models 460, determining scores 465, and saving the determined scores 470.”, Col. 26 Claim 1, "selecting, from the plurality of candidate predictive machine learning models generated at the server, a predictive machine learning model that satisfies a threshold level of accuracy."
Asher automatically evaluates each candidate model based on its predictive accuracy and selects the most accurate model. This automatic accuracy-based evaluation and selection of a single best model corresponds to automatically evaluating the plurality of candidates to generate a recommended re-trained ML model.)
outputting … prediction results and the recommended [trained] ML model to the computing system associated with the end user (Col. 6 Lines 5-13 of Asher, "The data center 120 may then transmit, for example to cloud client 105—an indication of a plurality of predicted values for the prediction field based in part on the predictive machine learning model.", Col. 9 Lines 2-5, "The database server 210 may transmit the plurality of predicted values by predicted value transmitter 265, by connection 220, to computing device 205.", Col. 26 Claim 1, "transmitting, to the cloud client for display in the user interface, an indication of a plurality of predicted values for the selected prediction field, wherein the plurality of predicted values are generated at the server using the selected predictive machine learning model."
Asher generates prediction results using the selected model at the cloud data center/server and transmits those prediction results to the end user's cloud-client computing device for display.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Saghir's cloud-based automated re-training of well-site ML models with Asher's generation of a plurality of candidate models, accuracy-based evaluation, and selection of a recommended model in order to identify and deliver the most accurate model and its prediction results to the end user (Paragraphs 4 and 33 of Asher).
Regarding claim 3,
Saghir teaches validating, via the cloud-based application, the trained ML model by determining whether the well prediction results are within predetermined accuracy thresholds (Paragraph 72 of Saghir, "Depending on the prediction accuracy requirement (i.e., required QoS), the training phase can be repeated until satisfactory results are achieved.", Paragraph 90, "whichever model or models produce the best (i.e., most accurate) results based on a predefined QoS are deployed at the edge device in a deployment phase 1210."
Saghir determines whether the model's prediction results satisfy a predefined accuracy requirement and repeats training until that threshold is met, which corresponds to validating the trained ML model by determining whether the well prediction results are within predetermined accuracy thresholds.)
Regarding claim 6,
Saghir teaches outputting, from the cloud-based application, a command signal to the computing system to automatically launch a well evaluation module on the computing system to present data relating to the well prediction results and the recommended re-trained ML model (Paragraph 7 of Saghir, "initiating a responsive action on the edge device based on the one or more dynagraph classifications. The responsive action includes at least one of logging a date and time, sending an alert message to a control system, adjusting a motor speed of a rod pump, or shutting off power to the rod pump…", Paragraph 80, "FIG. 7 shows an exemplary graphical presentation 700 of the outputs of the Ensemble model (or models) as displayed, for example, on an HMI…", Paragraph 87, "the resulting conclusions of the modeling (e.g., dynagraph classifications) are outputted at 1116, typically in terms of probabilities. This step can entail displaying the conclusions on an HMI of the edge device…"
Saghir initiates a responsive action including a signal to the control system that causes the edge computing system to present the modeling outputs on its HMI. The signal causing the edge device to present the classification/prediction data corresponds to a command signal to automatically launch a well evaluation module to present data relating to the well prediction results and the re-trained ML model.)
Regarding claim 8,
Saghir teaches [a] cloud-based infrastructure, comprising: a logging and control system configured to (Paragraph 61 of Saghir, "A control unit 128 at the well site 100 gathers data about various aspects of the well site 100 for monitoring and tracking purposes.", Paragraph 62, "an external system, such as a supervisory control and data acquisition (SCADA) system 134", Paragraph 118, "a Cloud platform"
Saghir's control unit, SCADA system, and Cloud platform correspond to the claimed logging and control system of a cloud-based infrastructure.)
train a machine learning (ML) model to create a trained ML model, wherein the trained ML model is configured to output well prediction results based on datasets specific to respective wells (Paragraph 68 of Saghir, "the one or more ML models 224 should be appropriately trained before the models can perform their functions effectively. Training may be done using a self-supervised learning method in which labeled historical data is applied as an input to the ML models.", Paragraph 59, "embodiments of the present disclosure provide methods and systems for developing and deploying robust ML modeling in an IIoT environment to automatically predict failures and non-optimal operations directly at the well site", Paragraph 110, "the data used to build this confusion matrix 2400 was obtained from 4 wells over a period of approximately 1 month."
Saghir trains one or more ML models on labeled well-site data to predict failures and non-optimal operations at the well site. The models are trained and applied on data obtained from respective wells, which corresponds to well prediction results based on datasets specific to respective wells.)
provide a data label module to a computing system associated with an end user (Paragraph 66 of Saghir, "A well site monitoring and control application 220, or rather the computer-readable instructions therefor, may also reside on or be downloaded to the storage device 210. The well site monitoring and control application 220… to perform ML-based analytics on dynagraphs from the RTU 130 to automatically detect abnormal operations and generate a response thereto.”, Paragraph 83, "For example, the ML models may be unable to identify a dynagraph card that was not included in the training phase if the card is unique to a specific pump behavior. As another example, the ML models may be unable to make a clear distinction between two dynagraph card types that are similar to each other, such as “gas interference” and “gas lock” dynagraph cards. In such cases, the dynagraph cards may be labeled locally by a field operator via a locally available HMI.", Paragraph 84, "In this example, the user interface 1000 includes several navigation buttons, indicated generally at 1002, that allow an operator to navigate and otherwise scroll through and access various functions available on the edge gateway. A dynagraph area 1004 displays the dynagraph of interest, which may be the current dynagraph or an older dynagraph…the operator may then label the dynagraph with his own classification, which is then stored in a feedback database with the dynagraph of interest for subsequent consideration and use in model retraining and readjustment."
Saghir provides the field operator's edge computing system with a monitoring and control application having a labeling interface (HMI/GUI 1000) by which the operator labels the data. This downloaded labeling application corresponds to the claimed data label module provided to a computing system associated with an end user.)
deploy the trained ML model to a cloud-based application, wherein the cloud-based application is configured to: (Paragraph 58, “ML-based analytics is implemented at well sites in conjunction with the broader movement toward a so-called Industrial Internet of Things (IIoT). In an IIoT landscape, well site operators increasingly make use of interconnected sensors and devices to collect and store large amounts of data directly at the well sites. This data can then be processed at the well sites by an edge device equipped with ML modeling to monitor and control well operations…The edge device can then generate alarms, carry out contingent actions, and store the results of the processing to the Cloud as needed. The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations.”, Paragraph 62 of Saghir, "the data may be forwarded to other systems within an enterprise 138 and/or to the Cloud 140 (which may include a private enterprise Cloud) for further processing as needed.", Paragraph 87, “In some embodiments, the outputs from the ML models are optionally provided to an Ensemble model at 1114 to produce more accurate results. In either case, the resulting conclusions of the modeling (e.g., dynagraph classifications) are outputted at 1116, typically in terms of probabilities. This step can entail displaying the conclusions on an HMI of the edge device, storing them in a local or remote storage device or database, sending them to an external system (e.g., a SCADA host), and/or uploading them to a Cloud-computing platform for further processing.”
Saghir deploys and manages the ML models via a Cloud platform that performs further processing, which corresponds to deploying the trained ML model to a cloud-based application.)
receive a training dataset specific to a well, wherein the training dataset comprises data labeling generated by the end user via the data label module (Paragraph 58, “The edge device can then generate alarms, carry out contingent actions, and store the results of the processing to the Cloud as needed. The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations.”, Paragraph 83 of Saghir, "the dynagraph cards may be labeled locally by a field operator via a locally available HMI. These manually labeled cards can then be used to retrain the models (or parts thereof)", Paragraph 84, "the operator may then label the dynagraph with his own classification, which is then stored in a feedback database with the dynagraph of interest for subsequent consideration and use in model retraining and readjustment.", Paragraph 118, "Alternatively, an edge gateway with ML-based analytics can take advantage of full Cloud connectivity where multiple such gateways can be maintained and managed via a Cloud platform. This option allows real-time data collected from multiple edge gateways to be used to retrain ML models"
The operator-labeled well-site cards are collected via the Cloud platform and used for retraining. The manually labeled dataset generated by the operator corresponds to a training dataset comprising data labeling generated by the end user via the data label module.)
re-train the trained ML model using the training dataset to generate a … re-trained ML model [candidate] (Paragraph 83 of Saghir, "These manually labeled cards can then be used to retrain the models (or parts thereof)", Paragraph 118, "Alternatively, an edge gateway with ML-based analytics can take advantage of full Cloud connectivity where multiple such gateways can be maintained and managed via a Cloud platform. This option allows real-time data collected from multiple edge gateways to be used to retrain ML models, resulting in more accurate applications at the edge. This retraining process can be automated and retrained models can be deployed to multiple gateways based on a preset schedule, whenever model accuracy reaches a predefined level, or when a unique dynagraph card is added (e.g., via Transfer Learning)."
Saghir re-trains the ML models on the Cloud using the newly labeled dataset in an automated retraining process. Saghir teaches the re-training via the cloud-based application using the training dataset to generate a single retrained ML model.)
output well prediction results and the … re-trained ML model to the computing system associated with the end user (Paragraph 80, “FIG. 7 shows an exemplary graphical presentation 700 of the outputs of the Ensemble model (or models) as displayed, for example, on an HMI (e.g., HMI 212) or other display of an edge gateway equipped with the ML-based analytics capability disclosed herein. The presentation 700 shows dynagraph classifications reflecting possible operating conditions for a given rod pump assembly over the course of one day based on the dynagraphs from that rod pump assembly, with the vertical axis indicating probability and the horizontal axis showing time.”, Paragraph 87 of Saghir, "In some embodiments, the outputs from the ML models are optionally provided to an Ensemble model at 1114 to produce more accurate results. In either case, the resulting conclusions of the modeling (e.g., dynagraph classifications) are outputted at 1116... This step can entail displaying the conclusions on an HMI of the edge device... sending them to an external system (e.g., a SCADA host), and/or uploading them to a Cloud-computing platform for further processing.", Paragraph 89, “the dynagraph classifications that are outputted at 1116 may be further processed and used to anticipate future pump failures. Thus, in addition to outputting the dynagraph classifications to a local or remote storage device or database, sending them to an external system, and so forth, the classifications and underlying dynagraphs may be further outputted or otherwise provided to one or more additional ML models on the edge device to anticipate future failures.”, Paragraph 118, "This retraining process can be automated and retrained models can be deployed to multiple gateways based on a preset schedule, whenever model accuracy reaches a predefined level, or when a unique dynagraph card is added (e.g., via Transfer Learning)."
Saghir outputs the prediction conclusions to the operator's edge device and deploys the retrained models back to the gateways, which corresponds to outputting the well prediction results and the re-trained ML model to the computing system associated with the end user.)
Saghir does not teach re-training, via the cloud-based application, the trained ML model using the training dataset to generate a plurality of re-trained ML model candidates or automatically evaluating the plurality of re-trained ML model candidates to generate a recommended re-trained ML model.
Asher, in the same field of endeavor, teaches [train]… the … ML model using the training dataset to generate a plurality of [trained] ML model candidates (Col. 8 Lines 38-43 of Asher, "the database server 210 may generate the predictive learning model based on training a plurality of candidate machine learning models. The plurality of candidate machine learning models may include any number of machine learning algorithms used in predictive model building.", Col. 10 Lines 28-33, “process flow 400 may also include receiving a selection of a plurality of fields 430, removing a subset of data 435, evaluation 440, removing one or more features 445, training learning models 450, evaluating learning models 455, selecting learning models 460, determining scores 465, and saving the determined scores 470.”, Col. 26 Claim 1, "generating a plurality of candidate predictive machine learning models at the server based at least in part on the plurality of features."
Asher trains, from the received data set, a plurality of candidate predictive machine learning models. Each such candidate model, when applied in Saghir's cloud-based re-training loop using the operator-labeled training dataset, corresponds to the claimed plurality of re-trained ML model candidates.)
automatically evaluate… the plurality of [trained] ML model candidates to generate a recommended [trained] ML model (Col. 8 Lines 43-50 of Asher, "The database server 210 may then evaluate the plurality of candidate machine learning models. This evaluation may be based in part on a predictive accuracy of each of the machines. The database server 210 may subsequently select the predictive machine learning model based in part on the evaluation (e.g., based on which model is most accurate, or is otherwise best suited for the selected data set or the desired predictive value).", Col. 10 Lines 28-33, “process flow 400 may also include receiving a selection of a plurality of fields 430, removing a subset of data 435, evaluation 440, removing one or more features 445, training learning models 450, evaluating learning models 455, selecting learning models 460, determining scores 465, and saving the determined scores 470.”, Col. 26 Claim 1, "selecting, from the plurality of candidate predictive machine learning models generated at the server, a predictive machine learning model that satisfies a threshold level of accuracy."
Asher automatically evaluates each candidate model based on its predictive accuracy and selects the most accurate model. This automatic accuracy-based evaluation and selection of a single best model corresponds to automatically evaluating the plurality of candidates to generate a recommended re-trained ML model.)
output … prediction results and the recommended [trained] ML model to the computing system associated with the end user (Col. 6 Lines 5-13 of Asher, "The data center 120 may then transmit, for example to cloud client 105—an indication of a plurality of predicted values for the prediction field based in part on the predictive machine learning model.", Col. 9 Lines 2-5, "The database server 210 may transmit the plurality of predicted values by predicted value transmitter 265, by connection 220, to computing device 205.", Col. 26 Claim 1, "transmitting, to the cloud client for display in the user interface, an indication of a plurality of predicted values for the selected prediction field, wherein the plurality of predicted values are generated at the server using the selected predictive machine learning model."
Asher generates prediction results using the selected model at the cloud data center/server and transmits those prediction results to the end user's cloud-client computing device for display.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Saghir's cloud-based automated re-training of well-site ML models with Asher's generation of a plurality of candidate models, accuracy-based evaluation, and selection of a recommended model in order to identify and deliver the most accurate model and its prediction results to the end user (Paragraphs 4 and 33 of Asher).
Claim 10 recites similar limitations to the limitations recited in claim 3. Therefore, claim 10 is rejected using the same rationale as claim 3.
Claim 13 recites similar limitations to the limitations recited in claim 6. Therefore, claim 13 is rejected using the same rationale as claim 6.
Claims 2, 9, 15, 16, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Saghir (US 20220316314 A1) in view of Asher (US 11663517 B2) and in further view of Whaley (US 10601786 B2).
Regarding claim 2,
Saghir teaches via the data label module… via a new data labeling module of the cloud-based application (Paragraph 83, "For example, the ML models may be unable to identify a dynagraph card that was not included in the training phase if the card is unique to a specific pump behavior. As another example, the ML models may be unable to make a clear distinction between two dynagraph card types that are similar to each other, such as “gas interference” and “gas lock” dynagraph cards. In such cases, the dynagraph cards may be labeled locally by a field operator via a locally available HMI.", Paragraph 84, "In this example, the user interface 1000 includes several navigation buttons, indicated generally at 1002, that allow an operator to navigate and otherwise scroll through and access various functions available on the edge gateway. A dynagraph area 1004 displays the dynagraph of interest, which may be the current dynagraph or an older dynagraph…the operator may then label the dynagraph with his own classification, which is then stored in a feedback database with the dynagraph of interest for subsequent consideration and use in model retraining and readjustment.", Paragraph 58, “ML-based analytics is implemented at well sites in conjunction with the broader movement toward a so-called Industrial Internet of Things (IIoT). In an IIoT landscape, well site operators increasingly make use of interconnected sensors and devices to collect and store large amounts of data directly at the well sites. This data can then be processed at the well sites by an edge device equipped with ML modeling to monitor and control well operations…The edge device can then generate alarms, carry out contingent actions, and store the results of the processing to the Cloud as needed. The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations.”);
Claim 9 recites similar limitations to the limitations recited in claim 2. Therefore, claim 9 is rejected using the same rationale as claim 2.
Regarding claim 15,
Saghir teaches [a] method, comprising: training a machine learning (ML) model to create a trained ML model, wherein the trained ML model is configured to output well prediction results based on datasets specific to respective wells (Paragraph 68 of Saghir, "the one or more ML models 224 should be appropriately trained before the models can perform their functions effectively. Training may be done using a self-supervised learning method in which labeled historical data is applied as an input to the ML models.", Paragraph 59, "embodiments of the present disclosure provide methods and systems for developing and deploying robust ML modeling in an IIoT environment to automatically predict failures and non-optimal operations directly at the well site", Paragraph 110, "the data used to build this confusion matrix 2400 was obtained from 4 wells over a period of approximately 1 month."
Saghir trains one or more ML models on labeled well-site data to predict failures and non-optimal operations at the well site. The models are trained and applied on data obtained from respective wells, which corresponds to well prediction results based on datasets specific to respective wells.)
providing a data label module to a computing system associated with an end user (Paragraph 66 of Saghir, "A well site monitoring and control application 220, or rather the computer-readable instructions therefor, may also reside on or be downloaded to the storage device 210. The well site monitoring and control application 220… to perform ML-based analytics on dynagraphs from the RTU 130 to automatically detect abnormal operations and generate a response thereto.”, Paragraph 83, "For example, the ML models may be unable to identify a dynagraph card that was not included in the training phase if the card is unique to a specific pump behavior. As another example, the ML models may be unable to make a clear distinction between two dynagraph card types that are similar to each other, such as “gas interference” and “gas lock” dynagraph cards. In such cases, the dynagraph cards may be labeled locally by a field operator via a locally available HMI.", Paragraph 84, "In this example, the user interface 1000 includes several navigation buttons, indicated generally at 1002, that allow an operator to navigate and otherwise scroll through and access various functions available on the edge gateway. A dynagraph area 1004 displays the dynagraph of interest, which may be the current dynagraph or an older dynagraph…the operator may then label the dynagraph with his own classification, which is then stored in a feedback database with the dynagraph of interest for subsequent consideration and use in model retraining and readjustment."
Saghir provides the field operator's edge computing system with a monitoring and control application having a labeling interface (HMI/GUI 1000) by which the operator labels the data. This downloaded labeling application corresponds to the claimed data label module provided to a computing system associated with an end user.)
deploying the trained ML model to a cloud-based application (Paragraph 58, “ML-based analytics is implemented at well sites in conjunction with the broader movement toward a so-called Industrial Internet of Things (IIoT). In an IIoT landscape, well site operators increasingly make use of interconnected sensors and devices to collect and store large amounts of data directly at the well sites. This data can then be processed at the well sites by an edge device equipped with ML modeling to monitor and control well operations…The edge device can then generate alarms, carry out contingent actions, and store the results of the processing to the Cloud as needed. The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations.”, Paragraph 62 of Saghir, "the data may be forwarded to other systems within an enterprise 138 and/or to the Cloud 140 (which may include a private enterprise Cloud) for further processing as needed.", Paragraph 87, “In some embodiments, the outputs from the ML models are optionally provided to an Ensemble model at 1114 to produce more accurate results. In either case, the resulting conclusions of the modeling (e.g., dynagraph classifications) are outputted at 1116, typically in terms of probabilities. This step can entail displaying the conclusions on an HMI of the edge device, storing them in a local or remote storage device or database, sending them to an external system (e.g., a SCADA host), and/or uploading them to a Cloud-computing platform for further processing.”
Saghir deploys and manages the ML models via a Cloud platform that performs further processing, which corresponds to deploying the trained ML model to a cloud-based application.)
receiving, via the cloud-based application, a training dataset specific to a well, wherein the training dataset comprises data labeling generated by the end user via the data label module (Paragraph 58, “The edge device can then generate alarms, carry out contingent actions, and store the results of the processing to the Cloud as needed. The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations.”, Paragraph 83 of Saghir, "the dynagraph cards may be labeled locally by a field operator via a locally available HMI. These manually labeled cards can then be used to retrain the models (or parts thereof)", Paragraph 84, "the operator may then label the dynagraph with his own classification, which is then stored in a feedback database with the dynagraph of interest for subsequent consideration and use in model retraining and readjustment.", Paragraph 118, "Alternatively, an edge gateway with ML-based analytics can take advantage of full Cloud connectivity where multiple such gateways can be maintained and managed via a Cloud platform. This option allows real-time data collected from multiple edge gateways to be used to retrain ML models"
The operator-labeled well-site cards are collected via the Cloud platform and used for retraining. The manually labeled dataset generated by the operator corresponds to a training dataset comprising data labeling generated by the end user via the data label module.)
re-training, via the cloud-based application, the trained ML model using the training dataset to generate a … re-trained ML model [candidate] (Paragraph 83 of Saghir, "These manually labeled cards can then be used to retrain the models (or parts thereof)", Paragraph 118, "Alternatively, an edge gateway with ML-based analytics can take advantage of full Cloud connectivity where multiple such gateways can be maintained and managed via a Cloud platform. This option allows real-time data collected from multiple edge gateways to be used to retrain ML models, resulting in more accurate applications at the edge. This retraining process can be automated and retrained models can be deployed to multiple gateways based on a preset schedule, whenever model accuracy reaches a predefined level, or when a unique dynagraph card is added (e.g., via Transfer Learning)."
Saghir re-trains the ML models on the Cloud using the newly labeled dataset in an automated retraining process. Saghir teaches the re-training via the cloud-based application using the training dataset to generate a single retrained ML model.)
outputting, from the cloud-based application, well prediction results and the … re-trained ML model to the computing system associated with the end user (Paragraph 80, “FIG. 7 shows an exemplary graphical presentation 700 of the outputs of the Ensemble model (or models) as displayed, for example, on an HMI (e.g., HMI 212) or other display of an edge gateway equipped with the ML-based analytics capability disclosed herein. The presentation 700 shows dynagraph classifications reflecting possible operating conditions for a given rod pump assembly over the course of one day based on the dynagraphs from that rod pump assembly, with the vertical axis indicating probability and the horizontal axis showing time.”, Paragraph 87 of Saghir, "In some embodiments, the outputs from the ML models are optionally provided to an Ensemble model at 1114 to produce more accurate results. In either case, the resulting conclusions of the modeling (e.g., dynagraph classifications) are outputted at 1116... This step can entail displaying the conclusions on an HMI of the edge device... sending them to an external system (e.g., a SCADA host), and/or uploading them to a Cloud-computing platform for further processing.", Paragraph 89, “the dynagraph classifications that are outputted at 1116 may be further processed and used to anticipate future pump failures. Thus, in addition to outputting the dynagraph classifications to a local or remote storage device or database, sending them to an external system, and so forth, the classifications and underlying dynagraphs may be further outputted or otherwise provided to one or more additional ML models on the edge device to anticipate future failures.”, Paragraph 118, "This retraining process can be automated and retrained models can be deployed to multiple gateways based on a preset schedule, whenever model accuracy reaches a predefined level, or when a unique dynagraph card is added (e.g., via Transfer Learning)."
Saghir outputs the prediction conclusions to the operator's edge device and deploys the retrained models back to the gateways, which corresponds to outputting the well prediction results and the re-trained ML model to the computing system associated with the end user)
Saghir does not teach re-training, via the cloud-based application, the trained ML model using the training dataset to generate a plurality of re-trained ML model candidates or automatically evaluating the plurality of re-trained ML model candidates to generate a recommended re-trained ML model.
Asher, in the same field of endeavor, teaches [training]… the … ML model using the training dataset to generate a plurality of [trained] ML model candidates (Col. 8 Lines 38-43 of Asher, "the database server 210 may generate the predictive learning model based on training a plurality of candidate machine learning models. The plurality of candidate machine learning models may include any number of machine learning algorithms used in predictive model building.", Col. 10 Lines 28-33, “process flow 400 may also include receiving a selection of a plurality of fields 430, removing a subset of data 435, evaluation 440, removing one or more features 445, training learning models 450, evaluating learning models 455, selecting learning models 460, determining scores 465, and saving the determined scores 470.”, Col. 26 Claim 1, "generating a plurality of candidate predictive machine learning models at the server based at least in part on the plurality of features."
Asher trains, from the received data set, a plurality of candidate predictive machine learning models. Each such candidate model, when applied in Saghir's cloud-based re-training loop using the operator-labeled training dataset, corresponds to the claimed plurality of re-trained ML model candidates.)
automatically evaluating… the plurality of [trained] ML model candidates to generate a recommended [trained] ML model (Col. 8 Lines 43-50 of Asher, "The database server 210 may then evaluate the plurality of candidate machine learning models. This evaluation may be based in part on a predictive accuracy of each of the machines. The database server 210 may subsequently select the predictive machine learning model based in part on the evaluation (e.g., based on which model is most accurate, or is otherwise best suited for the selected data set or the desired predictive value).", Col. 10 Lines 28-33, “process flow 400 may also include receiving a selection of a plurality of fields 430, removing a subset of data 435, evaluation 440, removing one or more features 445, training learning models 450, evaluating learning models 455, selecting learning models 460, determining scores 465, and saving the determined scores 470.”, Col. 26 Claim 1, "selecting, from the plurality of candidate predictive machine learning models generated at the server, a predictive machine learning model that satisfies a threshold level of accuracy."
Asher automatically evaluates each candidate model based on its predictive accuracy and selects the most accurate model. This automatic accuracy-based evaluation and selection of a single best model corresponds to automatically evaluating the plurality of candidates to generate a recommended re-trained ML model.)
outputting … prediction results and the recommended [trained] ML model to the computing system associated with the end user (Col. 6 Lines 5-13 of Asher, "The data center 120 may then transmit, for example to cloud client 105—an indication of a plurality of predicted values for the prediction field based in part on the predictive machine learning model.", Col. 9 Lines 2-5, "The database server 210 may transmit the plurality of predicted values by predicted value transmitter 265, by connection 220, to computing device 205.", Col. 26 Claim 1, "transmitting, to the cloud client for display in the user interface, an indication of a plurality of predicted values for the selected prediction field, wherein the plurality of predicted values are generated at the server using the selected predictive machine learning model."
Asher generates prediction results using the selected model at the cloud data center/server and transmits those prediction results to the end user's cloud-client computing device for display.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Saghir's cloud-based automated re-training of well-site ML models with Asher's generation of a plurality of candidate models, accuracy-based evaluation, and selection of a recommended model in order to identify and deliver the most accurate model and its prediction results to the end user (Paragraphs 4 and 33 of Asher).
Saghir in view of Asher does not teach encrypting the training dataset via a data label module or decrypting the training dataset via a new data labeling module without exposing the training dataset.
Whaley, in the same field of endeavor, teaches encrypting… the training dataset prior to sending the training dataset to the [database] (Paragraph 5 of Whaley, "During operation, the system receives encrypted sensor data at a gateway server, wherein the encrypted sensor data includes a client identifier corresponding to an associated user or client device."; Paragraph 16, "Client 111... uses symmetric key 113 to encrypt sensor data 112 (step 312). Next, the encrypted sensor data 112 is loaded into database 104 (step 314)."
Whaley's client encrypts the sensor data with a symmetric key before it is uploaded. The client, which generates and handles the data to be used for training, corresponds to the data label module encrypting the training dataset.)
and decrypting… the training dataset during re-training of the trained ML model without exposing the training dataset external to the new data labeling module and the trained ML model (Paragraph 16, "the encrypted sensor data 112 is moved inside secure enclave 108 and is decrypted using symmetric key 113 (step 316)... the anonymized sensor data 116 is sent to learner 107 for processing…", Paragraph 10, "any memory and code running inside the enclave cannot be read or written, even by the highest permission user of the system, the system author, or the owner of the system"; Paragraph 5, "neither the operator of the system, nor owner of the gateway server has access to the sensor data", Paragraph 3, “ML systems generally operate by training a model to perform a recognition task based on features extracted from large volumes of user-related data. The ML systems then use the trained model to perform the specific recognition task.”
Whaley's secure enclave decrypts the data and communicates it to the learner for training, and the data cannot be read outside the enclave, even by the system owner. The training dataset is decrypted during training without being exposed external to the enclave or the model.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to combine Saghir in view of Asher’s cloud re-training of confidential well data with Whaley's client-side encryption and secure-enclave decryption in order to preserve the privacy and confidentiality of the training data during re-training (Abstract and Paragraph 10 of Whaley).
Claim 16 recites similar limitations to the limitations recited in claim 3. Therefore, claim 16 is rejected using the same rationale as claim 3.
Claim 19 recites similar limitations to the limitations recited in claim 6. Therefore, claim 19 is rejected using the same rationale as claim 6.
Claims 4 and 11 are rejected under 35 U.S.C. 103 as being unpatentable over Saghir (US 20220316314 A1) in view of Asher (US 11663517 B2), in further view of Li-Bland (US 20220318689 A1).
Regarding claim 4,
Saghir teaches automatically evaluating, via the cloud-based application, the trained ML model using the training dataset to generate a … re-trained ML model [candidate] based on [the locally generated data] … between the well prediction results and related observation data. (Paragraph 58, “ML-based analytics is implemented at well sites in conjunction with the broader movement toward a so-called Industrial Internet of Things (IIoT). In an IIoT landscape, well site operators increasingly make use of interconnected sensors and devices to collect and store large amounts of data directly at the well sites. This data can then be processed at the well sites by an edge device equipped with ML modeling to monitor and control well operations….”, Paragraph 83 of Saghir, "These manually labeled cards can then be used to retrain the models (or parts thereof)", Paragraph 118, "Alternatively, an edge gateway with ML-based analytics can take advantage of full Cloud connectivity where multiple such gateways can be maintained and managed via a Cloud platform. This option allows real-time data collected from multiple edge gateways to be used to retrain ML models, resulting in more accurate applications at the edge. This retraining process can be automated and retrained models can be deployed to multiple gateways based on a preset schedule, whenever model accuracy reaches a predefined level, or when a unique dynagraph card is added (e.g., via Transfer Learning).", Paragraph 83, “As another example, the ML models may be unable to make a clear distinction between two dynagraph card types that are similar to each other, such as “gas interference” and “gas lock” dynagraph cards. In such cases, the dynagraph cards may be labeled locally by a field operator via a locally available HMI. These manually labeled cards can then be used to retrain the models (or parts thereof) to improve the ability of the models to identify a dynagraph card that reflects behavior that is unique to a particular pump or to better distinguish between similar cards.”, Col. 10 Lines 12-17, “The indication may be displayed at prediction display 350. In other examples, the database server may transmit a confidence prediction 335, which may indicate to a user a confidence level associated with the prediction.”
Saghir re-trains the ML models on the Cloud using the newly labeled dataset in an automated retraining process. Saghir teaches the re-training via the cloud-based application using the training dataset to generate a single retrained ML model.)
Saghir does not teach automatically evaluating… the plurality of re-trained ML model candidates to generate the recommended retrained ML model using a spider diagram analysis.
Asher, in the same field of endeavor, teaches automatically evaluating… the plurality of [trained] ML model candidates to generate the recommended [trained] ML model based on a results confidence index… between the … prediction results and related observation data. (Col. 8 Lines 43-50 of Asher, "The database server 210 may then evaluate the plurality of candidate machine learning models. This evaluation may be based in part on a predictive accuracy of each of the machines. The database server 210 may subsequently select the predictive machine learning model based in part on the evaluation (e.g., based on which model is most accurate, or is otherwise best suited for the selected data set or the desired predictive value).", Col. 26 Claim 1, "selecting, from the plurality of candidate predictive machine learning models generated at the server, a predictive machine learning model that satisfies a threshold level of accuracy.", Col. 3 Lines 27-33, “the database server may generate a predictive learning model based on training a plurality of candidate machine learning models. For example, the database server may evaluate a plurality of machine learning models based in part on a predictive accuracy of the model. The database may then, for example, select the predictive machine learning model based in part on the evaluation.” Col. 12, Lines 12-16, “For example, the score may be based on a number of predicted values or a correlation between a particular predicted value and the data set (e.g., the accuracy of the prediction).”
Asher trains a plurality of candidate models, evaluates each on its predictive accuracy, and selects the candidate that satisfies a threshold level of accuracy, and further outputs a confidence level associated with the resulting prediction. Selecting the recommended candidate on the basis of its accuracy and associated prediction confidence corresponds to generating the recommended re-trained ML model based on a results confidence index.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Saghir's cloud-based re-training of well-site ML models with Asher's accuracy- and confidence-based evaluation and selection of a candidate model in order to identify and deliver to the end user the candidate whose predictions are most accurate and carry the highest confidence (Col. 3 Lines 27-33 and Col. 8 of Asher).
Saghir in view of Asher does not teach generat[ing] the recommended re-trained ML model based on a results confidence index, a robustness to noise index, and a matching index between the well prediction results and related observation data.
Li-Bland, in the same field of endeavor, teaches generat[ing] the … ML model based on … a robustness to noise index and a matching index between [distributions of ML models] (Paragraph 47, “Moreover, the scoring functions are often noisy (e.g., random variables), have non-trivial correlations (since they are all correlated with the more fundamental but somewhat intractable notion of model quality), and/or not directly comparable, in the sense that they each may have different scales. It can be beneficial to have a method that combines the information provided in the values of these various scoring functions in order to select the best model in a way that handles the noise, variation, and/or correlation in the scores appropriately.”, Paragraph 2, “selection of an optimal machine learning (ML) model according to one or more heterogeneous and noisy metrics of model quality.”, Paragraph 37, “the scoring functions, unaltered by including a duplicate scoring function (in the noiseless setting), and insensitive to noisy measurements. An example of a process for selecting models in accordance with an embodiment of the invention is conceptually illustrated in FIG. 1.”
Li-Bland selects the optimal model from among a plurality of candidate models according to multiple metrics of model quality and combines those metrics so that the selection handles the noise in the scores and is not sensitive to noisy measurements. Evaluating the candidates and selecting the recommended model in a manner that is expressly robust to noise and is based off correlation in the quality scores corresponds to generating the recommended re-trained ML model based on a robustness to noise index and a matching index.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Saghir in view of Asher's evaluation and selection of a recommended re-trained model candidate with Li-Bland's noise-robust combination of multiple model-quality metrics in order to reliably select the best candidate when the quality scores are noisy and must be balanced against one another, rather than relying on a single noisy accuracy score (Paragraphs 3 and 47 of Li-Bland).
Claim 11 recites similar limitations to the limitations recited in claim 4. Therefore, claim 11 is rejected using the same rationale as claim 4.
Claims 5 and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Saghir (US 20220316314 A1) in view of Asher (US 11663517 B2), in view of Li-Bland (US 20220318689 A1), and in further view of Salonidis (US 20210065048 A1).
Regarding claim 5,
Saghir teaches automatically evaluating, via the cloud-based application, the trained ML model using the training dataset to generate a … re-trained ML model [candidate] (Paragraph 58, “ML-based analytics is implemented at well sites in conjunction with the broader movement toward a so-called Industrial Internet of Things (IIoT). In an IIoT landscape, well site operators increasingly make use of interconnected sensors and devices to collect and store large amounts of data directly at the well sites. This data can then be processed at the well sites by an edge device equipped with ML modeling to monitor and control well operations…The edge device can then generate alarms, carry out contingent actions, and store the results of the processing to the Cloud as needed. The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations.”, Paragraph 83 of Saghir, "These manually labeled cards can then be used to retrain the models (or parts thereof)", Paragraph 118, "Alternatively, an edge gateway with ML-based analytics can take advantage of full Cloud connectivity where multiple such gateways can be maintained and managed via a Cloud platform. This option allows real-time data collected from multiple edge gateways to be used to retrain ML models, resulting in more accurate applications at the edge. This retraining process can be automated and retrained models can be deployed to multiple gateways based on a preset schedule, whenever model accuracy reaches a predefined level, or when a unique dynagraph card is added (e.g., via Transfer Learning)."
Saghir re-trains the ML models on the Cloud using the newly labeled dataset in an automated retraining process. Saghir teaches the re-training via the cloud-based application using the training dataset to generate a single retrained ML model.)
Saghir does not teach automatically evaluating… the plurality of re-trained ML model candidates to generate the recommended retrained ML model using a spider diagram analysis.
Asher, in the same field of endeavor, teaches automatically evaluating… the plurality of [trained] ML model candidates to generate a recommended [trained] ML model (Col. 8 Lines 43-50 of Asher, "The database server 210 may then evaluate the plurality of candidate machine learning models. This evaluation may be based in part on a predictive accuracy of each of the machines. The database server 210 may subsequently select the predictive machine learning model based in part on the evaluation (e.g., based on which model is most accurate, or is otherwise best suited for the selected data set or the desired predictive value).", Col. 10 Lines 28-33, “process flow 400 may also include receiving a selection of a plurality of fields 430, removing a subset of data 435, evaluation 440, removing one or more features 445, training learning models 450, evaluating learning models 455, selecting learning models 460, determining scores 465, and saving the determined scores 470.”, Col. 26 Claim 1, "selecting, from the plurality of candidate predictive machine learning models generated at the server, a predictive machine learning model that satisfies a threshold level of accuracy."
Asher automatically evaluates each candidate model based on its predictive accuracy and selects the most accurate model. This automatic accuracy-based evaluation and selection of a single best model corresponds to automatically evaluating the plurality of candidates to generate a recommended re-trained ML model.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Saghir's cloud-based automated re-training of well-site ML models with Asher's generation of a plurality of candidate models, accuracy-based evaluation, and selection of a recommended model in order to identify and deliver the most accurate model and its prediction results to the end user (Paragraphs 4 and 33 of Asher).
Saghir in view of Asher does not teach generat[ing] a recommended [trained] ML model using a spider diagram analysis.
Salonidis, in the same field of endeavor, teaches to generate a recommended [trained] ML model using a spider diagram analysis (Paragraph 3, “The one or more machine learning composition modules generate one or more machine learning models. A machine learning model pipeline, an ensemble of a plurality of machine learning model pipelines, or a combination thereof, along with corresponding metadata, may be generated using the one or more machine learning composition modules.”, Paragraph 79, “the metadata may be placed as a node… on the radial UI visualization 470 that include multiple concentric rings. The concentric rings of the radial UI visualization 470 may include 1) pipeline rings 475 containing all available model pipelines and pipeline ensembles…”, Paragraph 80, “The radial UI visualization 470 may include a radial leaderboard. That is, pipelines may be sorted on different ranking criteria…the user …can visualize in a concise way, how machine learning model pipelines… are ranked according to different criteria.”, Paragraph 99, “the visualization operations may refine top performing nodes (e.g., pipeline 1 and pipeline 2 of FIG. 5A-5C …) along the way until the top number of selected performers … is determined. Once a top number of performers are selected, another layer/arc may be added outside of the previous arc.”
Salonidis places the plurality of candidate machine learning model pipelines as nodes on a radial visualization of concentric rings and sorts/ranks them on that radial "leaderboard" to visualize and refine the candidates down to the top-performing model. This radial concentric-ring visualization of candidate models corresponds to a spider diagram.)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Saghir in view of Asher's automated evaluation and selection of a recommended model candidate with Salonidis's radial spider-diagram leaderboard in order to allow the end user to visualize in a concise way how the candidate models rank across criteria and identify the top-performing model to recommend (Paragraphs 13 and 80 of Salonidis).
Claim 12 recites similar limitations to claim 5. Therefore, claim 12 is rejected using the same rationale as claim 5.
Claims 7 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Saghir (US 20220316314 A1) in view of Asher (US 11663517 B2) and in further view of Benslimane ("Automated Corrosion Analysis with Prior Domain Knowledge-Informed Neural Networks", June 2022).
Regarding claim 7,
Saghir in view of Asher does not teach wherein the well prediction results comprise well integrity prediction results.
Benslimane, in the same field of endeavor, teaches wherein the well prediction results comprise well integrity prediction results (Page 1 Introduction, “Pipe condition evaluation for well integrity workflows involves corrosion characterization. leak and structural deficiency monitoring, and detection of potential breaks in the pipe. Accurate corrosion detection is important for both operators and service companies to decrease interpretation time, unify monitoring subjectivity, and increase overall performance… Our main motivation in this paper is to propose 31 automated and robust workflow for corrosion and wear detection of steel pipes.” Page 9 Conclusion, "he detected defects are clustered into fine-grained categories and used to compute burst pressure indicators and provide valuable inputs to decision makers related to the well integrity."). Benslimane applies a deep-learning model to ultrasonic pipe data to detect corrosion and analyze casing thickness for computing burst-pressure and collapse indicators used in well-integrity decisions. These corrosion and casing-integrity prediction outputs correspond to well prediction results comprising well integrity prediction results.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to combine Saghir in view of Asher's cloud-based continual ML for well-site prediction with Benslimane's application of machine learning to well-integrity (corrosion and casing) evaluation in order to monitor pipe condition over the productive life of the well and support integrity decision-making (Introduction of Benslimane).
Claim 14 recites similar limitations to the limitations recited in claim 7. Therefore, claim 14 is rejected using the same rationale as claim 7.
Claim 17 is rejected under 35 U.S.C. 103 as being unpatentable over Saghir (US 20220316314 A1) in view of Asher (US 11663517 B2), in view of Whaley (US 10601786 B2 in further view of Li-Bland (US 20220318689 A1).
Claim 17 recites similar limitations to the limitations recited in claim 4. Therefore, claim 17 is rejected using the same rationale as claim 4.
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Saghir (US 20220316314 A1) in view of Asher (US 11663517 B2), in view of Whaley (US 10601786 B2), in view of Li-Bland (US 20220318689 A1), and in further view of Salonidis (US 20210065048 A1).).
Claim 18 recites similar limitations to the limitations recited in claim 5. Therefore, claim 18 is rejected using the same rationale as claim 5.
Claim 20 is rejected under 35 U.S.C. 103 as being unpatentable over Saghir (US 20220316314 A1) in view of Asher (US 11663517 B2), in view of Whaley (US 10601786 B2), and in further view of Benslimane ("Automated Corrosion Analysis with Prior Domain Knowledge-Informed Neural Networks", June 2022).
Claim 20 recites similar limitations to the limitations recited in claim 7. Therefore, claim 20 is rejected using the same rationale as claim 7.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MAJD MAHER HADDAD whose telephone number is (571)272-2265. The examiner can normally be reached Mon-Friday 8-5 pm.
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, Kamran Afshar, can be reached at (571) 272-7796. 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.
/M.M.H./Examiner, Art Unit 2125
/KAMRAN AFSHAR/Supervisory Patent Examiner, Art Unit 2125
