DETAILED ACTION
This non-final office action is responsive to application 18/404,159 as submitted 04 Jan. 2024.
Claim status is currently pending and under examination for claims 1-20 of which independent claims are 1, 8 and 15.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Priority
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119(e) or under 35 U.S.C. 120, 121, 365(c), or 386(c) is acknowledged. The application’s effective filing date is 01/06/2023.
Information Disclosure Statement
As required by MPEP 609(c), the applicant’s submissions of the Information Disclosure Statement dated 08/26/24 is acknowledged by the examiner and the cited references have been considered in the examination of the claims now pending. As required by MPEP 609 C(2), a copy of the PTOL-1449 initialed and dated by the examiner is attached to the instant office action.
Specification
The title of the invention is not descriptive. A new title is required that is clearly indicative of the invention to which the claims are directed, see MPEP 606.01.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. In determining whether the claims are subject matter eligible, the examiner applies guidance set forth under MPEP 2106.
Step 1: Is the claim to a process, machine, manufacture, or composition of matter? Yes—all claims fall within one of the four statutory categories: claims 1-7 are a method/process, claims 8-14 are a computer readable medium/article of manufacture, and claims 15-20 are a system/machine. As such, all claims are to statutory subject matter and the analysis should proceed per MPEP 2106.03.
Step 2A, prong one: Does the claim recite an abstract idea, law of nature or natural phenomenon? Yes—the claims, under the broadest reasonable interpretation, recites an abstract idea. In this case, claims fall within the enumerated grouping of abstract idea being “Mental Processes” under MPEP 2106.04(a)(2)(II), but for the recitation of generic computer components. More particularly, the claims recite:
“identifying, in response to receiving the predictions, a subset of the class of scenarios that are beyond a threshold tolerance of accuracy” (Mental observation/identifying with evaluation of acceptable performance)
“based on identifying the subset of the class of scenarios, generating […] data set that includes emphasized event data from a plurality of historical sporting events, wherein the emphasized event data emphasizes events associated with the subset of the class of scenarios” (Mental judgment to emphasize sporting event data set for subsequent learning, e.g. feature selection of choice data, or attention to info relevant to sports gambling. Generated mentally or manually by-hand with aid of pen and paper or template)
“identifying weights” (Mental observation, e.g. recognizing coefficients)
Focus of the claims emphasize sporting events as generated data based on identified class beyond threshold accuracy. This is similar to wagering games which have been found abstract by the courts. A person whom identifies and generates data through such functions may include a coach or scout for players looking at records or statistics of gameplay. As such, the claims recite at least mental processes as the abstract idea.
Step 2A, prong two: Does the claim recite additional elements that integrate the judicial exception into a practical application? No—a practical application is not integrated by the judicial exception because the additional elements are as follows:
“receiving, from a machine learning model, predictions associated with a class of scenarios” MPEP 2106.05(g) adding insignificant extra-solution activity to the judicial exception, e.g. mere data gathering from necessary model output
“generating, by the computing system, an updated machine learning model by:” MPEP 2106.05(f)(h) adding the words ‘apply-it’ with the judicial exception, merely uses a computer as a tool to perform an abstract idea, and/or generally linking the use of a judicial exception to a particular technological environment or field of use
“initializing the updated machine learning model using the weights” MPEP 2106.05(h) generally linking the use of the judicial exception to a particular technological environment or field of use
“training the updated machine learning model using the training data set, wherein training the updated machine learning model comprises modifying the weights based on the training data set” MPEP 2106.05(h) generally linking the use of the judicial exception to a particular technological environment or field of use
“deploying, by the computing system, the updated machine learning model” MPEP 2106.05(f)(g) adding the words ‘apply-it’ with the judicial exception, merely uses a computer as a tool to perform an abstract idea, and/or adding insignificant extra-solution activity to the judicial exception as post-solution
Balance of the claim concerns machine learning model update generating by computing system with initializing and training for receiving predictions and deploying model. This amounts to applying established techniques that generally convey the field of machine learning without meaningful limitation beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception under MPEP 2106.05(e). Further, MPEP 2106.04(a)(2)(III) “A claim that requires a computer may still recite a mental process.” As such, the claim remains drawn to mental processes as the abstract idea and the additional elements fail to integrate the judicial exception into a practical application.
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception? No—the claims do not include additional elements that amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea in to a practical application, the additional elements are identified with respect to MPEP 2106.05 and do not demonstrate an inventive concept.
“receiving, from a machine learning model, predictions associated with a class of scenarios” MPEP 2106.05(g) adding insignificant extra-solution activity to the judicial exception, e.g. mere data gathering from necessary model output. For example, a classifier predicts class and such functionality being conventional under MPEP 2106.05(d) is e.g. Zhang et al., US PG Pub No 2021/0004625A1 at [0094] “common classifiers include a softmax classifier”
“generating, by the computing system, an updated machine learning model by:” MPEP 2106.05(f)(h) adding the words ‘apply-it’ with the judicial exception, merely uses a computer as a tool to perform an abstract idea, and/or generally linking the use of a judicial exception to a particular technological environment or field of use. Particularly, the computing system is recited at a high level of generality and does not qualify as a particular machine under MPEP 2106.05(b), and the update may be generated per specification [0032] “any suitable type of machine learning” e.g. [0101] “not limited to a deep learning network”
“initializing the updated machine learning model using the weights” MPEP 2106.05(h) generally linking the use of the judicial exception to a particular technological environment or field of use. The initialization conveys a mere first and such feature being conventional is per Alabdulmohsin (below) at [0035] “conventional weight initialization”
“training the updated machine learning model using the training data set, wherein training the updated machine learning model comprises modifying the weights based on the training data set” MPEP 2106.05(h) generally linking the use of the judicial exception to a particular technological environment or field of use, e.g. [0031] “Any suitable type of training” which includes conventional per Valliappan et al., US PG Pub No 2023/0229958A1 as per [0013] “conventional ML system is trained and then deployed”
“deploying, by the computing system, the updated machine learning model” MPEP 2106.05(f)(g) adding the words ‘apply-it’ with the judicial exception, merely uses a computer as a tool to perform an abstract idea, and/or adding insignificant extra-solution activity to the judicial exception as post-solution. Particularly, said extra-solution activity is a conventional activity under MPEP 2106.05(d) per Valliappan at [0013] as noted above.
Significantly more is not satisfied by the balance of the claim for at least the above reasons. Using a computer/machine to perform machine learning according to the limitations as claimed fails to impart particularity through a technical solution that elevates the claim as a whole beyond performing routine optimization with established techniques. If the claim language provides only a result-oriented solution, with insufficient detail for how a computer accomplishes it, then the claims do contain an inventive concept. Taken alone, their additional elements do not amount to significantly more than the above-identified judicial exception (the abstract idea). Looking at the limitations as a whole, taken together as an ordered combination adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements satisfies the test of particular transformation. Their collective functions merely provide conventional computer implementation. Therefore, claim 1 is found ineligible for patent under 35 U.S.C. 101.
Independent claim 8 recites limitations similar to claim 1 and further recites “A non-transitory computer readable medium configured to store process-readable instructions, wherein when executed by a processor, the instructions perform operations.” These are considered additional elements which fall under MPEP 2106.05(f) mere instructions to implement an abstract idea on a computer, or merely uses a computer to perform an abstract idea. Particularly, the additional elements are recited at a high level of generality and do not qualify as a particular machine under MPEP 2106.05(b). Accordingly, the claim remains drawn to the abstract idea and the additional elements do not integrate the judicial exception into a practical application or amount to significantly more.
Independent claim 15 recites limitations similar to claim 1 and further recites “A system comprising: a processor; and a non-transitory computer readable medium having program instructions stored thereon, which, when executed by the processor, cause the system to perform operations.” These are considered additional elements which fall under MPEP 2106.05(f) mere instructions to implement an abstract idea on a computer, or merely uses a computer to perform an abstract idea. Particularly, the additional elements are recited at a high level of generality and do not qualify as a particular machine under MPEP 2106.05(b). Accordingly, the claim remains drawn to the abstract idea and the additional elements do not integrate the judicial exception into a practical application or amount to significantly more.
Dependent claims 2, 9 and 16 disclose receiving updated associated affiliated with class from the updated machine learning model and outputting a visualization of the updated predictions. The limitations are considered as additional elements which fall under MPEP 2106.05(g) adding insignificant extra-solution activity to the judicial exception. For example, gathering data of necessary model outputs for display in any format on a common computer screen. This is similar to displaying certain results of analysis which has been found by the courts as insufficient as per MPEP 2106.05(h). Accordingly, the claims remain drawn the judicial exception and the additional elements are insufficient to integrate the judicial exception into a practical application or amount to significantly more.
Dependent claims 3, 10 and 17 disclose identifying an additional subset of class scenarios beyond threshold accuracy and generating additional data that includes emphasized additional event data from sporting events that emphasizes the additional subset. This is considered part of the abstract idea which embellishes the identifying and generating data to include additional data. The limitations may be performed as mental observation and generating data manually by-hand or mentally as it does not preclude mental performance. There are no additional elements.
Dependent claims 4, 11 and 18 disclose sequentially exposing the updated model to boosts of training comprising the training data and additional training data. The limitation is considered as an additional element which amounts to adding insignificant extra-solution activity under MPEP 2106.05(g). Particularly, said extra-solution activity is conventional under MPEP 2106.05(d) as evidenced by Dong et al., US PG Pub No 2018/0005130A1 at [0037] “conventional XGboost classifier” Thus, the claim remains drawn to the judicial exception and the additional elements fail to integrate the judicial exception into a practical application or amount to significantly more.
Dependent claims 5, 12 and 19 disclose receiving indication from user that model predictions are beyond threshold accuracy. This is considered part of the abstract idea being mental processes of evaluation by human user indicating acceptable performance. There are no additional elements.
Dependent claims 6, 13 and 20 disclose providing first and second inputs to model for generating predictions and determining that the first and second predictions are within and outside of respective ranges. The determining limitations are considered to be part of the abstract idea including mental determinations for evaluating estimations as in and out of ranges. No indication is provided from the specification as to the criticality of ranges. The limitations of providing inputs to machine learning models for generating predictions is considered additional elements which fall under MPEP 2106.05(h) generally linking the use of the judicial exception to a particular technological environment or field of use. Merely providing input to generate output predictions renders the model a black-box without any hint of particular transformation or meaningful limitation. Therefore, the claim remains drawn to the judicial exception and the additional elements are insufficient to integrate the judicial exception into a practical application or amount to significantly more.
Dependent claim 7 and 14 disclose providing initial training data and subclass to an edge event machine learning model for receiving emphasized event data output, and wherein emphasized event data is a subset of the initial training data weighted higher than initial training data. The limitation of wherein emphasizing by weighting data higher than initial data is a judgment which can be mental process to embellish the abstract idea. The limitations of providing data and subclass to model for output is considered as additional elements which fall under MPEP 2106.05(h) generally linking the use of the judicial exception to a particular technological environment or field of use. In other words, providing input to receive output, by trained ML model, does not distill what transformation occurs within the model to produce the desired result. Without technical solution to distinguish the claimed invention from established practices in the field of machine learning, no inventive concept is found. Accordingly, the claim remains drawn to the abstract idea and the additional elements are insufficient to integrate the abstract idea into a practical application or amount to significantly more.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 8 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over:
Gultekin et al., US PG Pub No 2020/0250223A1 hereinafter Gultekin, in view of
Alabdulmohsin et al., US PG Pub No 2022/0253694A1 hereinafter Alabdulmohsin.
With respect to claim 1, Gultekin teaches:
A method {Gultekin [0090] “methods employed” e.g. Figs 3-5} comprising:
receiving, from a machine learning model, predictions associated with a class of scenarios {Gultekin Fig 4:450 “Predicted Class (Perception Tag)” received at 460, from model 440 in loop, similar at Fig 3:330 “Predicted Tags” training loop, scenario is shown Fig 5 football prediction, see e.g. [0131], [0079], and [0077] “class/tag”};
identifying, in response to receiving the predictions, a subset of the class of scenarios that are beyond a threshold tolerance of accuracy {Gultekin discloses [0126] “class-specific thresholds until a desired threshold of predictive accuracy is achieved” similar [0078] “sufficiently accurate predictions” to Fig 5 “Identify” with Fig 4 perception identifier model 410 and classifier 470. See also [0150] minimizing false positives, [0072-73]};
based on identifying the subset of the class of scenarios, generating, by a computing system, a training data set that includes emphasized event data from a plurality of historical sporting events, wherein the emphasized event data emphasizes events associated with the subset of the class of scenarios {Gultekin Fig 3 training perception model for Fig 5 football perception, see [0057] “generates a set of tags for each training sample (e.g., each image) of each class {e.g., each category of images}” is generating tags for training samples thus generated training data set comprising the tags based on class, for [0009] “sporting event” particularly football [0131] “predicting additional perception tags within the reduced domain of this ‘football’ context… generates the ‘NFL’ 532 perception tag, indicating that the target content is an NFL football game-a further reduction in contextual domain of the target content” with events comprising e.g. [0134] “actions such as passing the ball, kicking the ball, etc.)”. See also [0072,84] “Training Service 130 employs Class-Specific Threshold Service 132 to submit the entire set of training samples images” and “Employing the class-specific thresholds produced in step 350, Perception Identifier 175 generates a set of core tags for each training sample”. Computing system is shown Fig 1};
generating, by the computing system, an updated machine learning model {Gultekin Fig 3:330 described [0079] “adjust model weights” is update for “training a neural network” using “Training Service 130” 130 is of computing system Fig 1. See also [0057], [0019]} by:
training the updated machine learning model using the training data set, wherein training the updated machine learning model comprises modifying the weights based on the training data set {Gultekin Fig 3:345,330 “Re-Train” for “Training (over multiple Epochs) – Forward/Backward Propagation to adjust Weights” re-training over iterations/epochs to adjust/modify weights is updating, the model as [0079] “training a neural network representing the Perception …Perception in step 330” similarly at [0069-70]}; and
Gultekin further suggests [0056] “initial training” e.g. [0068] “must be trained initially”.
However, Gultekin does not appear to fairly teach or suggest the following limitations which are disclosed by Alabdulmohsin:
identifying weights of the machine learning model {Alabdulmohsin see Fig 2:202 “Identify current values of the weights” for 208 “neural network” described e.g. [0047] “identifies current values of the weight of the plurality of neural network layers as of a given training time step (step 202)”},
initializing the updated machine learning model using the weights {Alabdulmohsin [0046] “initializes the values of the weight of the layers of the neural network and performs an initial training round to update” and [0055] “re-initializes the values of the weights of at least the neural network”. See also [0006] “repeatedly re-initializing the values of the weights”, Figs 1:130, 2:208, 3}, and
deploying, by the computing system, the updated machine learning model {Alabdulmohsin [0043] “deploys the trained neural network 110 on one or more computing devices” e.g. per [0086] “deployed using a machine learning framework, e.g., a TensorFlow framework” Fig 1}.
Alabdulmohsin is directed to machine learning model training thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to identify and initialize weights for model deployment per Alabdulmohsin to arrive at the invention as claimed as applying known techniques to known methods ready for improvement to yield predictable results and/or for a motivation being [0006] “by repeatedly re-initializing the values of the weights or ‘higher’ layers in the neural network during training, the system can train a network to have improved generalization… result in improved accuracy” similar at [0065], and because the deployed model may be used to provide inference and perform machine learning tasks on edge devices [0043].
With respect to claim 8, the rejection of claim 1 is incorporated. The difference in scope being a non-transitory computer medium configured to store instructions executable by processor to perform limitations of method claim 1. Gultekin discloses [0051-54] “software 114 to implement desired functionality for the training and/or use of models… hardware and software 122, such as processors and memory (with processors executing instructions stored in volatile and/or non-volatile memory)” similar at [0044-45], and shown Fig 1. The remainder of this claim is rejected for the same rationale as claim 1.
With respect to claim 15, the rejection of claim 1 is incorporated. The difference in scope being a system comprising processor and non-transitory computer readable medium storing instructions executable by processor to perform limitations of method claim 1. Gultekin Fig 1 shows a system of server and clients, described [0051-53] “hardware and software components 112, such as processors and memory (with processors executing instructions stored in volatile and/or non-volatile memory)” similar at [0044-45] and/or [0121-23] “GPU servers to re-train”. The remainder of this claim is rejected for the same rationale as claim 1.
Claims 2, 9 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Gultekin and Alabdulhohsin in view of Ruiz et al., US PG Pub No 2019/0228290A1 hereinafter Ruiz.
With respect to claim 2, the combination of Gultekin and Alabdulmohsin teaches the method of claim 1. Gultekin discloses re-training for predictions Fig 3 with display using API to receive content [0130,114]. However, Gultekin does not explicitly state “updated predictions.” Ruiz teaches further comprising:
receiving updated predictions, associated with the class of scenarios, from the updated machine learning model {Ruiz [0128] “updated predicted outcome” received by GUI Figs 5, 6:606-10, class from machine learning model comprises random forest classifier [0142, 0050], iterative adjustment for training is disclosed [0101], and scenarios regard “sporting events” [0004] described throughout}; and
outputting a visualization of the updated predictions {Ruiz [0128] “GUI 530 that reflects the adjusted starting lineup with an updated predicted outcome” [0125-134]}.
Ruiz is directed to machine learning models for sporting events thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to specify updated predictions for display per Ruiz in combination with Gultekin’s re-trained model and API in combination to arrive at the invention as claimed as obvious to try in choosing from a finite number of identified, predictable solution, with a reasonable expectation of success to provide uses with display of new or updated predictions, and/or a motivation [0133] “generate GUI 550 that includes graphical representations of the adjusted lineup and the new predicted outcome” thus providing users with current estimate of relevant information with an up-to-date display for user-friendly feedback or interaction.
With respect to claim 9, the combination of Gultekin and Alabdulmohsin teaches the non-transitory computer readable medium of claim 8, and further combination with Ruiz teaches limitation of claim 2. Therefore, the rejection of claim 2 with equal motivation is applied to claim 9.
With respect to claim 16, the combination of Gultekin and Alabdulmohsin teaches the system of claim 15, and further combination with Ruiz teaches limitation of claim 2. Therefore, the rejection of claim 2 with equal motivation is applied to claim 16.
Claims 3-4, 10-11 and 17-18 are rejected under 35 U.S.C. 103 as being unpatentable over Gultekin, Alabdulmohsin and Ruiz in view of Soleymani et al., “Progressive Boosting for Class Imbalance” hereinafter Soleymani (arXiv: 1706.01531v1).
With respect to claim 3, the combination of Gultekin, Alabdulmohsin and Ruiz teaches the method of claim 2. Soleymani teaches further comprising:
identifying, from the updated predictions, an additional subset of the class of scenarios that are beyond the threshold tolerance of accuracy {Soleymani [P.10 Sect.3] “positive and negative class” class imbalance described where “the error of the classifier is determined based on its ability to correctly classify” correctness to comprise [P.19 ¶3] “performance metrics that can be maximized to set the decision threshold are accuracy” teaches an accuracy threshold, e.g. [P.12 ¶1] “accuracy criterion of 0.5 in AdaBoost” Alg.1 LineV, Alg.2 LineVIII. See also [P.6 ¶2] “subset of negative class…positive class”}; and
based on identifying the additional subset of the class of scenarios, generating, by the computing system, an additional training data set that includes emphasized additional event data from the plurality of historical sporting events, wherein the emphasized additional event data emphasizes the additional subset of the class of scenarios {Soleymani discloses [P.14 ¶2 - 15 ¶4] “positive and negative classes are generated… training data generated”, emphasized by weighting per Eqs. 26-27 [P.11] W+ and W- are weighting (W) positive (+) and negative (-) classes, applied at [P.13] Alg.2 LineVII within the For-loop Line5, and which may comprise [P.12 ¶2] “assigning higher weights” The dataset used suggests applicability to sport events as [P.13 ¶1] “Face in Action (FIA) video database …face re-identification application”}.
Soleymani is directed to training of classifiers thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to identify and generate with emphasis on class per Soleymani in combination with Gultekin’s sporting events to arrive at the invention as claimed for a motivation of addressing class imbalance so-titled, [P.1 ¶1] “Class imbalance is a fundamental issue in many real-world pattern application applications” with contributions helping to [P.3 ¶3] “avoid bias of performance towards the negative class”, and further [P.25 ¶2] “improve the classification accuracy.”
With respect to claim 4, the combination of Gultekin, Alabdulmohsin, Ruiz and Soleymani teaches the method of claim 3, wherein training the updated machine learning model using the training data set comprising:
sequentially exposing the updated machine learning model to boosts of training data comprising the training data set and the additional training data set {Soleymani discloses [P.3 ¶1] “During each Boosting iteration, a new base classifier is trained” further detailed [P.10 Sect.3] “Progressive Boosting… In each iteration, a subset of this temporary set is used for training such that the most important samples plus samples from the new partition are given an equally high opportunity to be used in training a base classifier” implemented Alg.2 [P.13] where iterations of a For-loop Line 5 for a vector process corresponds to sequentially, see e.g. Fig 2 and noting video data as time-series/sequence [P.13 ¶1], [P.15 ¶6]. See also [P.20 ¶4] “Each iteration of Boosting ensembles includes a validation step that should be added to training”}.
A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to boost training data sequentially per Soleymani for a motivation “Boosting procedure to avoid losing information while generating a diverse pool of classifiers” [Abst], [P.10 Sect.3 ¶2].
With respect to claim 10, the combination of Gultekin, Alabdulmohsin and Ruiz teaches the non-transitory computer readable medium of claim 9, and further combination with Soleymani teaches limitation of claim 3. Therefore, the rejection of claim 3 with equal motivation is applied to claim 10.
With respect to claim 11, the combination of Gultekin, Alabdulmohsin, Ruiz and Soleymani teaches the non-transitory computer readable medium of claim 10, and further teaches limitation of claim 4. Therefore, the rejection of claim 4 with equal motivation is applied to claim 11.
With respect to claim 17, the combination of Gultekin, Alabdulmohsin and Ruiz teaches the system of claim 16, and further combination with Soleymani teaches limitation of claim 3. Therefore, the rejection of claim 3 with equal motivation is applied to claim 17.
With respect to claim 18, the combination of Gultekin, Alabdulmohsin, Ruiz and Soleymani teaches the system of claim 17, and further teaches limitation of claim 4. Therefore, the rejection of claim 4 with equal motivation is applied to claim 18.
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Gultekin and Alabdulhohsin in view of Jana et al., US PG Pub No 2024/0144081A1 hereinafter Jana.
With respect to claim 5, the combination of Gultekin and Alabdulmohsin teaches the method of claim 1. Jana teaches wherein identifying the class of scenarios comprises:
receiving an indication from a user that the machine learning model is generating predictions beyond the threshold tolerance of accuracy {Jana [0061] “user-specified or desired accuracy threshold (e.g., performance metric(s)) with respect to predictions for custom classes” see e.g. Figs 3-5}.
Jana is directed to machine learning model training thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to specify accuracy threshold for classes per Jana in combination to arrive at the invention as claimed for a motivation of “improved accuracy” [0060] and which “affords the user more control over the training” [0077], [0062].
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Gultekin and Alabdulhohsin in view of Koda et al., PCT WO2024/042714A1 hereinafter Koda.
With respect to claim 6, the combination of Gultekin and Alabdulmohsin teaches the method of claim 1, wherein identifying the class of scenarios comprises:
providing a first set of inputs to the machine learning model to generate a first prediction {Gultekin [0079] “training samples into a format for use in training a neural network…input” for “predicted ‘tag’ outputs” is output prediction from neural network/model input, see Figs 3-4};
providing a second set of inputs to the machine learning model to generate an additional prediction {Gultekin [0081] “predicted probability with respect to each of the training samples” [0072] “predicted probability with respect to each of the training sample images associated with that class”};
However, Gultekin in combination does not disclose expected ranges which is taught by Koda:
determining that the first prediction is within an expected range of a first expected prediction {Koda [Yellow Highlight] “calculates IND data (in-distribution data) that is within the range of training data of the DNN model” where “E represents the expected value”}; and
determining the second prediction is outside of a second expected range of an second expected prediction {Koda [Green Highlight] “calculation unit 102 uses OOD (out-of-distribution data) that is outside the range of training data for the DNN model” and “E represents the expected value”}.
Koda is directed to neural network training thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to specify ranges as the in-distribution and out-of-distribution ranges per Koda in combination to arrive at the invention as claimed as applying a known technique to a known method ready for improvement to yield predictable results and/or a motivation of an [Orange Highlight] “allowable range of performance” based on “combining the first importance Lm(IND) and the second importance Lm (OOD)”.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Gultekin and Alabdulhohsin in view of Cheng et al., US PG Pub No 2023/0055636A1 hereinafter Cheng.
With respect to claim 7, the combination of Gultekin and Alabdulmohsin teaches the method of claim 1. Cheng teaches wherein the emphasized event data is generated by:
providing initial training data used to train the machine learning model to an edge event machine learning model {Cheng discloses [0094] “Slowfast neural network model…initialized with weights using a training dataset” being [0119] “trained with candidate clips containing goals extracted from games in the train set” so as for [0126] “detecting target events (which may also be referred to as events of interest or actions)” and/or [0082] “rare events”};
providing the subset of the class of scenarios to the edge event machine learning model {Cheng [0094] “Slowfast neural network…The network may be finetuned (1310) as a classifier” where “feature extractors are used to classify 4-second clips in to 4 categories” category/classes thus provided to the network/model by feature extraction, see similarly at [0144,152] and/or [0104-05]}; and
receiving the emphasized event data output by the edge event machine learning model based on the initial training data and the subset of the class of scenarios {Cheng [0153-59] “output may be computed as a weighted sum of the values… classifications are used (2520) to update the action recognition model” emphasized by weighting to include self-attention [0132] “Self-attention mechanism is employed to capture long-range context information and dynamically adjust weights according to the input”},
Chen is directed to generative model training for sporting events thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to provide training data and class for output by edge event model per Cheng in combination for motivation to [0051] “boost the performance of event-of-interest spotting” and further gives [0077] “three reasons this approach is preferred… contain more context information” cont’d “second reason is robustness… third, by analyzing shorter clips for the event of interest rather than the entire video, many resources (processing, processing time, memory, energy consumption, etc.) are preserved.”
Cheng suggests [0158] “weighted blend (e.g., two inputs blended 60% and 40%)” i.e. 60 > 40
However, Cheng does not expressly disclose “weighted higher” which is disclosed by Soleymani:
wherein the emphasized event data is a subset of the initial training data weighted higher than in the initial training data {Soleymani [P.12 ¶2] “assigning higher weights” as “More importance is given to these samples by assigning higher weights to them, so that they have a higher chance to be included in the training subset(s)”}.
Soleymani is directed to training of classifiers thus being analogous. A person having ordinary skill in the art would have considered it obvious prior to the effective filing date to assign higher weights per Soleymani to arrive at the invention as claimed for as applying known techniques to know methods ready for improvement to yield predictable results and/or a motivation of importance sampling for inclusion into the training set [P.12 ¶2].
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Gultekin, Alabdulhohsin, Ruiz and Soleymani in view of Jana.
With respect to claim 12, the combination of Gultekin, Alabdulhohsin, Ruiz and Soleymani teaches the non-transitory computer readable medium of claim 10, and further combination with Jana teaches the limitation of claim 5. Therefore, the rejection of claim 5 with equal motivation is applied to claim 12.
Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over Gultekin, Alabdulhohsin, Ruiz and Soleymani in view of Koda.
With respect to claim 13, the combination of Gultekin, Alabdulmohsin, Ruiz and Soleymani teaches the non-transitory computer readable medium of claim 10, and further combination with Koda teaches the limitation of claim 6. Therefore, the rejection of claim 6 with equal motivation is applied to claim 13.
Claim 14 is rejected under 35 U.S.C. 103 as being unpatentable over Gultekin, Alabdulhohsin, Ruiz and Soleymani in view of Cheng.
With respect to claim 14, the combination of Gultekin, Alabdulhohsin, Ruiz and Soleymani teaches the non-transitory computer readable medium of claim 10, and further combination with Chen and Soleymani teaches the limitation of claim 7. Therefore, the rejection of claim 7 with equal motivation is applied to claim 14.
Claim 19 is rejected under 35 U.S.C. 103 as being unpatentable over Gultekin, Alabdulhohsin and Ruiz in view of Jana.
With respect to claim 19, the combination of Gultekin, Alabdulhohsin and Ruiz teaches the system of claim 16, and further combination with Jana teaches the limitation of claim 5. Therefore, the rejection of claim 5 with equal motivation is applied to claim 19.
Claim 20 is rejected under 35 U.S.C. 103 as being unpatentable over Gultekin, Alabdulhohsin and Ruiz in view of Koda.
With respect to claim 20, the combination of Gultekin, Alabdulmohsin and Ruiz teaches the system of claim 16, and further combination with Koda teaches the limitation of claim 6. Therefore, the rejection of claim 6 with equal motivation is applied to claim 20.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Deliege et al., “SoccerNet-v2: A Dataset and Benchmarks for Holistic Understanding of Broadcast Soccer Videos” arXiv: 2011.13367v3 see Figs 1,5
Cioppa et al., “Scaling up SoccerNet with multi-view spatial localization and re-identification” discloses SoccerNet-v3, see Figs 1, 4
Zhu et al., “A Transformer-based System for Action Spotting in Soccer Videos” Figs 1-4
Bhanu et al., US PG Pub No 2020/0394413A1 trained GAN, classifier for athlete skills
Katz et al., US PG Pub No 2018/0082152A1 hockey video object detection
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Chase P Hinckley whose telephone number is (571)272-7935. The examiner can normally be reached M-F 9:00 - 5:00.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Miranda M. Huang can be reached at 571-270-7092. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/CHASE P. HINCKLEY/Examiner, Art Unit 2124
Prosecution Insights