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 .
DETAILED ACTION
The instant application having Application No. 17192057 has a total of 15 claims pending in the application, of which claims 2-3, 8, and 10-11 have been cancelled.
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, 4-7, 9, and 12-15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1: Claim 1 is a machine type claim. Claim 9 is a process type claim. Therefore, claims 1, 4-7, 9, and 12-15 are directed to either a process, machine, manufacture or composition of matter.
As per claims 1 and 9
2A Prong 1:
“Generate a prediction model for predicting an event” The user mentally creates some form of model to predict a future event by looking at data and making predictions based on that data.
“Generate a plurality of prediction models using the plurality of first training data, to thereby generate a prediction model for calculating a final predicted value based on predicted values of the plurality of prediction models” The user mentally or with pencil and paper sets up various prediction models based upon the data.
“Apply a plurality of … algorithms to the respective plurality of first training data, to thereby generate a plurality of first level prediction models, each first level prediction model corresponding to the respective different … algorithms” The user mentally or with pencil and paper creates a bunch of first level prediction models using the appropriate training data and algorithms.
“Execute each of the first level prediction models to generate predicted values from each of the first level prediction models” The user mentally or with pencil and paper executes the models to get results.
“calculate meta-features from each of the generated predicted values of the plurality of first level prediction models, and select optimal meta-features among the calculated meta-features to optimize prediction accuracy” The user mentally or with pencil and paper determines the best meta-features to use to improve the models.
“generate second training data including a plurality of sample data using the optimal meta-features calculated from each of the generated predicted values of the plurality of first level prediction models, and the correct prediction value for the event” The user mentally or with pencil and paper gathers up additional data on the performance of the models
“apply the plurality of different …algorithms to the second training data, to thereby generate a plurality of second level prediction models each corresponding to the respective different machine learning algorithms for outputting a plurality of final predicted values of the event” The user mentally or with pencil and paper uses the new data to create another model.
“Evaluate each of the final predicted values and determine an optimal … algorithm to be used for the second level prediction model based on the evaluation” The user mentally or with pencil and paper judges which algorithm is the best, and uses it.
“wherein the plurality of first training data include: training data for generating the prediction models in which global features of the event is reflected” The user mentally or with pencil and paper identifies the global feature within the data.
“training data for generating the prediction models in which local features of the event is reflected, each local feature indicating specific conditions of one or more of the global feature” The user mentally or with pencil and paper identifies the local features within the data.
2A Prong 2: This judicial exception is not integrated into a practical application.
Additional elements:
“A computer system”, “The computer system”, “at least one computer”, “an arithmetic device”, “a storage device”, “a connection interface”, “A storage unit”, (mere instructions to apply the exception using a generic computer component);
“a plurality of different machine learning algorithms”, “the respective different machine learning algorithms”, “machine learning algorithm”, “an optimal machine learning algorithm”, (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: The claims require nothing but a generic unnamed machine learning algorithm and the generic training of that algorithm with no additional detail. This could be any off the shelf machine learning algorithm);
“A storage unit configured to store a plurality of first training data including a plurality of sample data including values of a plurality of feature variables and a prediction correct value of the event” (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g)).
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Additional elements:
“A computer system”, “The computer system”, “at least one computer”, “an arithmetic device”, “a storage device”, “a connection interface”, “A storage unit”, (mere instructions to apply the exception using a generic computer component)
“a plurality of different machine learning algorithms”, “the respective different machine learning algorithms”, “machine learning algorithm”, “an optimal machine learning algorithm”, (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: The claims require nothing but a generic unnamed machine learning algorithm and the generic training of that algorithm with no additional detail. This could be any off the shelf machine learning algorithm);
“A storage unit configured to store a plurality of first training data including a plurality of sample data including values of a plurality of feature variables and a correct prediction value of the predicted event” (MPEP 2106.05(d)(II) indicate that merely “storing and retrieving information in memory” is a well‐understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed storing step is well-understood, routine, conventional activity is supported under Berkheimer).
As per claims 7, and 15, this claim includes additional mental steps similar to claim 1, and is rejected for similar reasons.
As per claims 4 and 12, this claim includes additional mental steps similar to claim 1, and is rejected for similar reasons.
2A Prong 2: This judicial exception is not integrated into a practical application.
Additional elements:
“Receive input data including a plurality of data including values of a plurality of variables, and information indicating the feature variables of the sample data included in the respective plurality of first training data” (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g)).
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Additional elements:
“Receive input data including a plurality of data including values of a plurality of variables, and information indicating the feature variables of the sample data included in the respective plurality of first training data” (MPEP 2106.05(d)(II) indicate that merely “receiving and transmitting data” is a well‐understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed receiving step is well-understood, routine, conventional activity is supported under Berkheimer).
As per claims 5 and 13, this claim includes additional mental steps similar to claim 1, and is rejected for similar reasons.
2A Prong 2: This judicial exception is not integrated into a practical application.
Additional elements:
“Receive input data including a plurality of data including values of a plurality of variables” (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g)).
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Additional elements:
“Receive input data including a plurality of data including values of a plurality of variables” (MPEP 2106.05(d)(II) indicate that merely “receiving and transmitting data” is a well‐understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed receiving step is well-understood, routine, conventional activity is supported under Berkheimer).
As per claims 6 and 14, this claim includes additional mental steps similar to claim 1, and is rejected for similar reasons.
2A Prong 2: This judicial exception is not integrated into a practical application.
Additional elements:
“Output the presentation information” (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g)).
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Additional elements:
“Output the presentation information” (MPEP 2106.05(d)(II) indicate that merely presenting data is a well‐understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed output step is well-understood, routine, conventional activity is supported under Berkheimer).
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.
Claims 1, 4-7, 9, and 12-15 are rejected under 35 U.S.C. 103 as being unpatentable over Xie et al (“A Combination of Boosting and Bagging for KDD Cup 2009 – Fast Scoring on a Large Database”) in view of Dirac et al (US 20150379072 A1), Zhou (“Ensemble Methods: Foundations and Algorithms”), Solomatine et al (“AdaBoost.RT: A Boosting Algorithm for Regression Problems”) and Bright et al (US 20110099122 A1).
As per claims 1 and 9, Xie discloses, “A … system configured to generate a prediction model” (Abstract; EN: This denotes the creation of multiple gradient boosted trees in order to make predictions). “for predicting an event, the … system comprising: ” (Abstract; EN: this denotes looking for churn (i.e. loss of customers), which meets the broadest reasonable interpretation of an event as it denotes the “event” of customers leaving a business).
“.. A plurality of first training data including a plurality of sample data values including values of a plurality of feature variables and a correct prediction value of the predicted event” (Pg.38-39, particularly section 3.1; EN: this denotes training data for the system, with the training set being the plurality of sample data values including features, and the prediction correct value being the correct label for those associated features).
“.. to generate a plurality of prediction models using the plurality of first training data” (Pg.40, particular section 3.4; EN: this denotes using bagging to create five boosted tree models from the training data). “To thereby generate a prediction model for calculating an ultimate predicted value based on predicted values of the plurality of prediction models, wherein” (Pg.40, particularly section 3.4; EN: this denotes averaging the outputs of all five models to give a final (“ultimate”) input).
“apply a plurality of … machine learning algorithms to the respective plurality of first training data, to thereby generate a plurality of first level prediction models” (Pg.39-40, particularly section 3.3; EN: this denotes creating the boosted trees, each which will consist of multiple algorithms. Here, the first algorithm of each tree is the plurality of machine learning algorithms that are used to create the “First level” prediction models, in this case the first machine learning model of each of the five boosted trees).
“execute each of the first level prediction models to generate predicted values from each of the first level prediction models” (Pg.39-40, particularly section 3.3; EN; this denotes getting results from the algorithms).
“Apply the plurality of … machine learning algorithms to the … training data, to thereby generate a … level prediction model for outputting a plurality of final predicted values of the event” (Pg.40, particularly section 3.4; EN: this denotes averaging the outputs of all five models to give a final input. When combined with the Solomatine reference, this denotes using the outputs of the various levels of the Boosting tree of the Xie reference to predict a final output).
“evaluate each of the final predicted values …” (pg.40, particularly section 3.4; EN: this denotes averaging the outputs for a result).
“wherein the plurality of first training data include:” (Pg.40, particular section 3.4; EN: this denotes using bagging to create five boosted tree models from the training data).
“training data for generating the prediction models in which global features of the event is reflected” (Pg.37, particularly the third paragraph; EN: this denotes “global” features such as numerical variables).
However, Xie fails to explicitly disclose, “a computer system”, “the computer system”, “at least one computer including an arithmetic device, a storage device, and a connection interface”, “a storage unit configured to store…”, “wherein the at least one computer is programmed to”, “different machine learning algorithms”, “each first level prediction model corresponding to the respective different machine learning algorithms”, “Calculate meta-features from each of the generated predicted values of the plurality of first level prediction models, and select optimal meta-features among the calculated meta-features to optimize prediction accuracy”, “Generate second training data including a plurality of sample data using the optimal meta-features calculated from each of the generated predicted values of the plurality of first level prediction models and the correct value of the prediction value of the event”, “Apply the plurality of different machine learning algorithms to the second training data, to thereby generate a plurality of second level prediction models each corresponding to the respective different machine learning algorithms …”, “evaluate each of the final predicted values and determine an optimal machine learning algorithm to be used for the second level prediction model based on the evaluation”
and “training data for generating the prediction models in which local features of the event is reflected, the local feature indicating specific conditions of one or more of the global features.”
Dirac discloses, “a computer system”, “the computer system” (Pg.10, particularly paragraph 0170; EN: this denotes the computing system to implement the machine learning system).
“at least one computer including an arithmetic device, a storage device, and a connection interface” (Pg.10, particularly paragraph 0170; EN: This denotes a processor, memory, and I/O interface).
“A storage unit configured to store…” (Pg.10, particularly paragraph 0170; EN: This denotes a memory used to store things as needed).
“wherein the at least one computer is programmed to” (Pg.10, particularly paragraph 0170; EN: This denotes a processor, memory, and I/O interface).
Zhou discloses, “different machine learning algorithms”, “each first level prediction model corresponding to the respective different machine learning algorithms”, “the plurality of different machine learning algorithms”, “… the respective different machine learning algorithms…” (Pg.15, particularly section 1.4; EN: this denotes the use of heterogeneous learners, ensemble learners that make use of different machine learning algorithms simultaneously to build the ensemble).
“evaluate each of the final predicted values and determine an optimal machine learning algorithm to be used for the second level prediction model based on the evaluation” (Pg.12-14, particularly section 1.3; EN: this denotes various ways of evaluating and selecting the best model to use).
Solomatine discloses, “Calculate meta-features from each of the generated predicted values of the plurality of first level prediction models, and select optimal meta-features among the calculated meta-features to optimize prediction accuracy” (Pg.1165, particularly C1, second to last paragraph; EN: this denotes determining error after the previous machine learning model based on a pre-set threshold, with error rates below that threshold being treated as correct classifications and those over the threshold being considered to be incorrect, and the incorrect classifications being used to train the next model, here the optimal meta-features are those that have a higher error rate than the threshold, as they are what it used to improve the model by training the next classifier).
“Generate second training data including a plurality of sample data using the optimal meta-features calculated from each of the generated predicted values of the plurality of first level prediction models and the correct value of the prediction value of the event” (Pg.1165, particularly C1, second to last paragraph; EN: this denotes determining error after the previous machine learning model based on a pre-set threshold, with error rates below that threshold being treated as correct classifications and those over the threshold being considered to be incorrect, and the incorrect classifications being used to train the next model).
“Apply the … machine learning algorithm to the second training data, to thereby generate … second level prediction model…” (Pg.1165, particularly C2, second paragraph; this details the creation of the next machine learning model based upon the error rate (meta-feature) of the previous model).
Bright discloses, “training data for generating the prediction models in which local features of the event is reflected, the local feature indicating specific conditions of one or more of the global features” (Pg.6, particularly paragraph 0054; EN: this denotes numerous different variables, all of which can be assigned to an individual. So each customer will have the combination of a particular measurement data, weight, sex, etc).
Xie and Dirac are analogous art because both involve machine learning.
Before the effective filing date it would have been obvious to one skilled in the art of machine learning to combine the work of Xie and Dirac in order to provide the required hardware to train multiple machine learning algorithms as needed.
The motivation for doing so would be to “implement one or more of the components of a machine learning service” (Dirac, Pg.8, paragraph 0170) or in the case of Xie, allow the system to train and create their machine learning models with appropriate computer equipment.
Therefore before the effective filing date it would have been obvious to one skilled in the art of machine learning to combine the work of Xie and Dirac in order to provide the required hardware to train multiple machine learning algorithms as needed.
Xie and Zhou are analogous art because both involve ensemble learning.
Before the effective filing date it would have been obvious to one skilled in the art of ensemble learning to combine the work of Xie and Zhou in order to use different algorithms as needed for ensemble learning.
The motivation for doing so would be to allow the use of “a common ensemble architecture…” (Zhou, Pg.15, second paragraph) to provide “the generalization ability of an ensemble if often much stronger than that of base learners” (Zhou, pg.15, third paragraph) or in the case of Xie, allow the system to use different machine learning algorithms as needed to produce their machine learning system.
Therefore before the effective filing date it would have been obvious to one skilled in the art of ensemble learning to combine the work of Xie and Zhou in order to use different algorithms as needed for ensemble learning.
Xie and Solomatine are analogous art because both involve boosting algorithms.
Before the effective filing date it would have been obvious to one skilled in the art of boosting algorithms to combine the work of Xie and Solomatine in order to use meta data from previous algorithms to create another algorithm.
The motivation for doing so would be to “Give enough attention to the examples with low values. Moreover, weight updating parameter is computed differently o give more emphasis on harder example when error rate is very low” (Solomatine, Pg.1164, C1, last paragraph) or in the case of Xie, allow the system to focus on the more difficult examples when performing a boosting algorithm and thus optimize for the issues the system is having trouble with.
Therefore before the effective filing date it would have been obvious to one skilled in the art of boosting algorithms to combine the work of Xie and Solomatine in order to use meta data from previous algorithms to create another algorithm.
Xie and Bright are analogous art because both involve machine learning.
Before the effective filing date it would have been obvious to one skilled in the art of machine learning to combine the work of Xie and Bright in order to allow combinations of variables to be grouped in training data.
The motivation for doing so would be to “relate variables in the training data (e.g. stored in columns in each row of the training data) to an objective variable” (Bright, Pg.7, paragraph 0062) or in the case of Xie, allow the system to use combinations of variables related to individuals, such as the customers of Xie, in order to consider how those combination of variables relate to the labels of the Xie reference.
Therefore before the effective filing date it would have been obvious to one skilled in the art of machine learning to combine the work of Xie and Bright in order to allow combinations of variables to be grouped in training data.
As per claims 4 and 12, Xie discloses, “receive input data including a plurality of data which includes values of a plurality of variables and information indicating the feature variables of the sample data including in the respective plurality of first training data”” (Pg.36-37, particularly section 2.1; EN: this denotes the system taking in all the variables and their various values).
“Generate the plurality of first training data from the input data on a basis of the information” (Pg.38-39, particularly sections 3.1; EN: this denotes the method of variable selection for training).
Dirac discloses, “wherein the at least one computer is programmed to:” (Pg.10, particularly paragraph 0170; EN: this denotes the computing system to implement the machine learning system).
As per claim 5 and 13, Xie discloses, “receive input data which includes a plurality of data including values of a plurality of variables” (Pg.36-37, particularly section 2.1; EN: this denotes the system taking in all the variables and their various values).
“analyze the plurality of variables of the data included in the input data” (Pg.36-37, particularly section 2.1; EN: this denotes the system taking in all the variables and their various values as well as analyzing them).
“Generate the plurality of first training data form the input data on a basis of a result of the analysis” (Pg.36-37, particularly section 2.1; EN: this denotes the system taking in all the variables and their various values as well as analyzing them. Pg.38-39, particularly sections 3.1; EN: this denotes the method of variable selection for training).
Dirac discloses, “wherein the at least one computer is programmed to:” (Pg.10, particularly paragraph 0170; EN: this denotes the computing system to implement the machine learning system).
As per claims 6 and 14, Xie discloses, “generate a prediction accuracy of the… prediction model” (Pg.39, particularly the third paragraph; EN: this denotes training and testing the model to check for performance (i.e. accuracy)).
“generate, based on a result of the evaluation of the prediction accuracies of the … prediction model, … information for … features to be used for training the … prediction model and a type of the machine learning algorithm to be applied to the … training data to that achieves a highest prediction accuracy…” (Pg.39, particularly the third paragraph; EN: this denotes training and testing the model to check for performance (i.e. accuracy). It further denotes testing different variables to find those that are most important, and selecting them to be used with their intended algorithm, the boosting tree (i.e. a type of machine learning algorithm)).
Dirac discloses, “wherein the at least one computer is programmed to:” (Pg.10, particularly paragraph 0170; EN: this denotes the computing system to implement the machine learning system).
Zhou discloses, “wherein the evaluation is an evaluation of the prediction accuracies of the final predicted values” and “… being the optimal machine learning model” (Pg.12-14, particularly section 1.3; EN: this denotes various ways of evaluating and selecting the best model to use).
Solomatine discloses “evaluate prediction accuracy of the second level prediction model”, “second level prediction model” and “meta-features” (Pg. 1165, particularly C2, third paragraph; EN: this denotes repeating the process to create multiple machines as needed).
However, Xie and Solomatine fail to explicitly disclose, “Presentation information for presenting a combination of the….”
Dirac discloses, “Presentation information for presenting a combination of the…” and “output the presentation information” (Pg.11, particularly paragraph 0085; EN: this denotes a user interface, including a GUI, in order to display information including features and other data).
Dirac and Xie modified by Solomatine are analogous art because both involve machine learning.
At the time of invention it would have been obvious to one skilled in the art of machine learning to further modify the already modified work of Xie and Solomatine in order to display the various features being manipulated in regards to the machine learning process.
The motivation for doing so would be to allow the system to “receive a request form a client via programmatic interface (such as an API, a command line tool, a web page, or a custom GUI) to perform a particular operation of an entity belonging to a set of supported entity types of the MLS” or in the case of Xie and Solomatine, allow the system to include a display in order to allow the user to understand what features are being used and why when manipulating and using these features to train the machine learning algorithm.
Therefore at the time of invention it would have been obvious to one skilled in the art of machine learning to further modify the already modified work of Xie and Solomatine in order to display the various features being manipulated in regards to the machine learning process.
As per claims 7 and 15, Dirac discloses, “wherein the at least one computer is programmed to:” (Pg.10, particularly paragraph 0170; EN: this denotes the computing system to implement the machine learning system).
Xie fails to explicitly disclose, “generate, as information to be used for prediction processing that is executed when data to be predicted is input, prediction processing pipeline information including details of processing for generating the first training data from the input data, details of processing for generating the second training data, and information on the second level prediction model.”
Dirac further discloses, “generate, as information to be used for prediction processing that is executed when data to be predicted is input, prediction processing pipeline information including details of processing for generating the first training data from the input data, details of processing for generating the second training data, and information on the second level prediction model” (Pg.10, particularly paragraph 0079; EN: this denotes performing the machine learning process in a pipeline, with the respective information coming from the Solomatine and Xie reference of the details of the actual algorithm).
Dirac and Xie modified by Solomatine are analogous art because both involve machine learning.
At the time of invention it would have been obvious to one skilled in the art of machine learning to further modify the already modified work of Xie and Solomatine in order to make use of a pipeline for the machine learning processing.
The motivation for doing so would be to allow the system to “implement a set of programmatic interface enabling such scheduled recurring operations…. A separately managed data pipelining service implemented as a provider network may be used in conjunction with the MLS for supporting such recurrent operations” (Dirac, Pg.10, paragraph 0079) or in the case of Xie and Solomatine, allow the machine learning service of the Dirac reference to be used to train and perform the various steps of the machine learning process as needed.
Therefore at the time of invention it would have been obvious to one skilled in the art of machine learning to further modify the already modified work of Xie and Solomatine in order to make use of a pipeline for the machine learning processing.
Response to Arguments
In pg.10, the Applicant argues in regards to the rejection under U.S.C. 101 of the independent claims,
When considering claim 1 as a whole, as required, the claim sets forth an improvement to increasing the prediction accuracy of prediction models when using training data having a variety of different feature variables. Further, claims 1 and 9 have been amended to set forth “calculate meta-features from each of the generated predicted values of the plurality of first level prediction models, and select optimal meta-features among the calculated meta-features to optimize prediction accuracy.” The optimal meta-features are then used to generate the second training data. This further prevents a reduction in the prediction accuracy. See “optimization 2” explained in the specification, which states:
In response, the Examiner maintains the rejection as shown above. The Applicant’s amendments denote improvements to the abstract idea, particularly selecting “optimal meta-features” is something that can be performed in the human mind mentally or with pencil and paper, and thus does not disclose an improvement to a technology. This causes the claim to be an abstract idea lacking significantly more, and therefore the rejection is maintained under U.S.C. 101.
In pg.12, the Applicant argues in regards to the rejection under U.S.C. 101,
In conventional technologies, such as described in JP-2013-164863-A, which employs ensemble learning, a variety in learning algorithms is provided, but a difference in features indicated by feature variables is not considered. See para. [0010]. Thus, a technical problem in the conventional technologies is that the features of events that are reflected in prediction models are biased. See id. The presently claimed invention solves this problem by evaluating combinations of first level prediction models, second training data based on the output of the first level prediction models and second level prediction models to determine the optimal machine learning algorithm to be used for a second level prediction model, which in turn improves the prediction accuracy when using training data having a variety of features, as explained above. See para. [0134].
In response, the Examiner maintains the rejection as shown above. Applicant describes the selection of the best features/meta-features for the training of generic machine learning algorithms to be an improvement to a technology. However, the selection of these features/meta-features is an abstract idea, and the machine learning algorithms are generic, machine learning algorithms that amount to no more than off the shelf machine learning algorithms. In improvement to the selection of data being used by machine learning models is an improvement to the abstract idea of selecting features, not to machine learning models, and therefore the rejection is maintained as shown above.
In pg.14, the Applicant argues in regards to the rejection under U.S.C. 101,
Further, the human mind is not equipped to and cannot practically apply a plurality of different machine learning algorithms to the respective plurality of first training data, to thereby generate a plurality of first level prediction models, execute each of the first level prediction models, or apply the plurality of different machine learning algorithms to the second training data, to thereby generate a plurality of second level prediction models each corresponding to the respective different machine learning algorithms for outputting a plurality of final predicted values of the event. Therefore at least these elements are additional elements to be considered under the Step 2A Prong Two analysis.
In response, the Examiner maintains the rejection as shown above. There is no attempt to describe the human mind performing a plurality of different machine learning models. This is handled under 2A/2B Prong2, which shows that using generic machine learning algorithms is not enough to be significantly more than the abstract idea. Therefore there is no attempt to state that performing machine learning/using different algorithms is performed in the human mind, this is part of the “apply it” aspects of the judicial exception to generic computer equipment or machine learning algorithms, and therefore the rejection is maintained as shown above.
In pg.15-16, the Applicant argues in regards to the rejection under U.S.C. 101,
For example, an inventive concept can be found in the non-conventional and non-generic arrangement of the features of the claims (e.g., the combination of features of the claims is non-conventional and non-generic). See BASCOM Global Internet Services, Inc., v. AT&T Mobility LLC, AT&T Corp., 827 F.3d 1341 (Fed. Cir. 2016); see also Ancora Techs. v. HTC Am., Inc., 908 F.3d 1343, 1346 (Fed. Cir. 2018) (determining that the claims at issue, relating to improving security against a computer's unauthorized use of a program, were directed to an improvement in computer functionality due to the claims having “the specificity required to transform [the] claim[s] from...claiming only a result to...claiming a way of achieving it” under Step 2A, but also acknowledging that such a technical improvement can establish eligibility under Step 2B in view of BASCOM). In BASCOM, the court indicated that, when looking at the claim as an ordered combination of claim limitations, “an inventive concept can be found in the non-conventional and non-generic arrangement of known, conventional pieces.” Similar to the concepts discussed in BASCOM, Applicant’s claim 1 includes additional elements that are sufficient to ensure that the claims amount to significantly more than an abstract idea.
In response, the Examiner maintains the rejection as shown above. The current claims do not contain a combination of features that are significantly more than the abstract idea. The claim calls for generic “different machine learning algorithms” with no details or limitations beyond generic, off the shelf machine learning algorithms. Otherwise the claim contains generic computer equipment such as arithmetic devices, storage devices, and connection interfaces which are standard to generic computer hardware with no particular organization or setup other than that they exist. Therefore the claims are not significantly more than the abstract idea and the 101 rejection is maintained as shown above.
Applicant's remaining arguments with respect to claims 1, 4-7, 9, and 12-15 have been considered but are moot in view of the new ground(s) of rejection or are mere repetitions of the above arguments and the rejections are maintained for the reasons described above.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BEN M RIFKIN whose telephone number is (571)272-9768. The examiner can normally be reached Monday-Friday 9 am - 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, Alexey Shmatov can be reached at (571) 270-3428. 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.
/BEN M RIFKIN/Primary Examiner, Art Unit 2123