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 .
Remarks
This Office Action is responsive to Applicants' Amendment filed on April 09, 2026, in which claims 1-5, 11, and 16 are amended. No claims have been newly added or cancelled. Claims 1-20 are currently pending.
Response to Arguments
With regards to the rejections of claims 1-20 under 35 U.S.C 101 for being directed towards abstract ideas, Applicant’s arguments have been considered, however Examiner respectfully disagrees. Applicant first argues that at least claim 1 is eligible at least under Step 2A, Prong 2 of the Subject Matter Eligibility Test at least by integrating any recited abstract ideas into a practical application. Applicant states on page 10 of the Remarks:
“Specifically, Applicant submits that the claims are directed to improvements in operations of data science models such as neural networks or decision trees on computer systems. The improvements in operations of the data science models improve the efficiency, performance, and adaptability of the data science models themselves and the computer systems operating the data science model.”
And Applicant further states on page 10 of the Remarks:
“The pending claims solve a technical problem of allowing the data science models to automatically and robustly compensate for deprecation of variables that no longer have data available without the need for user-implemented retraining or reforming of the entire data science model”.
Applicant further elaborates that claim 1 recites the technical solution of the invention by reciting limitations that adapt the determination of a risk assessment decision to depreciation of variables in a decision tree. Applicant finishes by citing paragraph [0034] of the instant application’s specification to show how the limitations recited in claim 1 provide the technical solution of the invention.
Examiner respectfully disagrees that claim 1 is eligible at Step 2A, Prong 2 of the Subject Matter Eligibility Test. Although Examiner does not dispute that the invention provides the described improvement, Examiner notes that MPEP 2106.04(d).III. states “Because a judicial exception alone is not eligible subject matter, if there are no additional claim elements besides the judicial exception, or if the additional claim elements merely recite another judicial exception, that is insufficient to integrate the judicial exception into a practical application”, i.e. that an improvement provided solely by improving an abstract idea itself is ineligible. Examiner further notes that most limitations within claim 1 have been identified as reflecting an abstract idea, as a decision tree model is simple and straightforward enough to be conceptualized and operated purely mentally by a human mind, or with the aid of pen and paper, including pruning (i.e. disregarding) branches of the decision tree, as well as making a risk assessment decision using a decision tree.
The only further limitations within claim 1 that are not identified as abstract ideas relate to providing or receiving data, which reflect mere extra-solution activity and do not integrate any recited judicial exceptions into a practical application themselves, as shown in the 101 rejection below.
Applicant further argues that claim 1 is also eligible at Step 2B of the Subject Matter Eligibility Test for reciting significantly more than any recited judicial exceptions. Applicant states on page 11 of the Remarks:
“the Office Action's asserted idea that ‘the additional limitations do not include significantly more than the underlying abstract idea; does not involve any of the various specific claim features discussed above, and those features impose meaningful limits on such a broad alleged idea. For example, the various features relating to pruning the decision tree after an intermediate node based on deprecation of at least one of the variables that no longer has accessible data and replacing the intermediate node with an output node that provides a decision result based on a majority of decision results made at the intermediate node before the pruning recite substantially more than an abstract idea”.
Examiner respectfully disagrees that claim 1 is eligible at Step 2B of the Subject Matter Eligibility Test. As stated previously, most of the limitations of claim 1 are identified as abstract ideas, as a decision tree can be created and used by the human mind purely mentally, or with the aid of pen and paper. Limitations identified as abstract ideas, which are judicial exceptions, are not considered at Step 2B, as stated in MPEP 2106.05.I.: “an ‘inventive concept’ is furnished by an element or combination of elements that is recited in the claim in addition to (beyond) the judicial exception”. The only additional limitations within claim 1 that are not identified as abstract ideas relate to providing or receiving data, which reflect mere extra-solution activity and do not integrate any recited judicial exceptions into a practical application themselves, as shown in the 101 rejection below.
With regards to the rejections of claims 1-6 and 9 under 35 U.S.C. 103 for being unpatentable over Huang et al. (U.S. Patent Pub. No. 2022/0035908) in view of Vinayagasundaram et al. “Efficient Gaussian Decision Tree Method for Concept Drift Data Stream”, with regards to the rejections of claims 11-15 under 35 U.S.C. 103 for being unpatentable over Huang, in view of Vinayagasundaram, in view of Wang, further in view of Barddal et al. “A survey on Feature Drift Adaptation”, and with regards to the rejections of claims 16-17, as being unpatentable over Huang, in view of Vinayagasundaram, further in view of Barddal, Applicant’s arguments have been considered but are moot in view of a new grounds of rejection of the claims, necessitated by Applicant’s amendment, as presented below.
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 abstract ideas without significantly more.
Regarding claim 1,
Step 1 - “Is the claim to a process, machine, manufacture or composition of matter?”
Yes, the claim is directed towards a process.
Step 2A, Prong 1 - “Is the claim directed to a law of nature, a natural phenomenon (product of nature) or an abstract idea?”:
The limitation of … determine a risk assessment decision for an operation associated with a user using a decision tree, wherein the decision tree includes a plurality of nodes interconnected by branches, the decision tree beginning with one or more input nodes and ending with a plurality of output nodes having decision results, recites an evaluation of risk and a decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of assessing, at the computer system, the dataset to determine at least one variable in the dataset that no longer has data existing in the dataset; recites an evaluation of variables and their data within a dataset, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining, at the computer system, the at least one variable in the dataset that no longer has data existing in the dataset to be at least one deprecated variable for the decision tree recites an evaluation of variables within a dataset, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of in response to determining the at least one deprecated variable in the dataset, pruning at least one branch in the decision tree, wherein the decision tree is pruned after an intermediate node based on the at least one deprecated variable in the dataset recites a judgement of what to prune within the tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of and wherein the intermediate node is replaced with an output node that provides a decision result based on a majority of decision results made at the intermediate node before the pruning recites an evaluation of pruning the tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining distinct decision results at the output nodes recites an evaluation of the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining a risk prediction based on a combination of the distinct decision results in the decision tree recites an evaluation of the result of the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of and determining, by the computer system, the risk assessment decision for the user based on the determined risk prediction for the user recites evaluation of the result of the risk prediction, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Step 2A, Prong 2 - “Does the claim recite additional elements that integrate the judicial exception into a practical application?”:
The limitation of receiving, by a computer system, a request to determine a risk assessment decision for an operation associated with a user…and wherein the request includes a dataset of variables associated with the user, the variables having corresponding nodes in the decision tree; recites the mere extra-solution activity of data gathering, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g).
The limitation of providing the dataset to a decision tree; recites the mere extra-solution activity of providing data to a decision tree for classification, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g).
Step 2B - “Does the claim recite additional elements that amount to significantly more than the judicial exception?”:
The limitation of receiving, by a computer system, a request to determine a risk assessment decision for an operation associated with a user…and wherein the request includes a dataset of variables associated with the user, the variables having corresponding nodes in the decision tree; recites receiving data over a network, which is well-understood, routine, and conventional, MPEP 2106.05(d).II.i.
The limitation of providing the dataset to a decision tree; recites providing data to a decision tree for classification, which is well-understood, routine, and conventional, (Kurama “An Introduction to Decision Trees” Pg. 1) “Decision trees are now widely used in many applications for predictive modeling, including both classification and regression”.
Therefore, claim 1 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 2,
Claim 2 adds the additional limitations to claim 1:
wherein the dataset of variables in the request has data for the at least one deprecated variable removed from the dataset after generation of the decision tree recites a judgement on what to include within the dataset of variables, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 2 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 3,
Claim 3 adds the additional limitation of wherein the intermediate node for the pruning is a node providing a decision result based on the at least one deprecated variable to claim 1, which recites mere additional details on the node that is pruned, without changing that in response to determining the at least one deprecated variable in the dataset, pruning at least one branch in the decision tree, wherein the decision tree is pruned after an intermediate node based on the at least one deprecated variable in the dataset, as recited in claim 1, is a judgement, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 3 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 4,
Claim 4 adds the additional limitations to claim 1:
deprecating the at least one variable in the dataset of variables in the request recites an evaluation of the request, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
wherein the at least one variable is deprecated based on changes in information available for determining the risk assessment decision recites an evaluation of the changes in information, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 4 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 5,
Claim 5 adds the additional limitations to claim 1:
wherein the decision tree includes a plurality of branches with intermediate nodes providing decision results based on the at least one deprecated variable recites mere additional details on the decision tree, without changing that providing the dataset to a decision tree, as recited in claim 1, is well-understood, routine, and convention extra-solution activity.
the method further comprising: pruning the plurality of branches in the decision tree, wherein the decision trees are pruned after the intermediate nodes providing decision results based on the at least one deprecated variable recites an evaluation of the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
and wherein the intermediate nodes are replaced with output nodes that provide decision results based on majorities of decision results made at the intermediate nodes before the pruning recites a judgement on how to prune the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 5 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 6,
Claim 6 adds the additional limitation of wherein pruning the at least one branch in the decision tree includes removing nodes that are downstream of the intermediate node on the pruned branch to claim 1, which recites mere additional details on the pruning, without changing that pruning at least one branch in the decision tree, wherein the decision tree is pruned after an intermediate node based on deprecation of at least one of the variables in the dataset that no longer has data accessible by the decision tree, as recited in claim 1, is a judgement, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 6 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 7,
Claim 7 adds the additional limitation of wherein the risk prediction is determined by averaging the distinct decision results in the decision tree to claim 1, which recites an evaluation of the distinct decision results of the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 7 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 8,
Claim 8 adds the additional limitation of wherein the risk prediction is determined by determining a majority decision result from the distinct decision results in the decision tree to claim 1, which recites an evaluation of the distinct decision results of the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 8 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 9,
Claim 9 adds the additional limitation of further comprising pruning, after a set of decision results, one or more branches in the decision tree that lack prediction power in the set of decision results to claim 1, which recites a judgement of what to prune, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 9 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 10,
Claim 10 adds the additional limitation of wherein the decision tree includes a random selection of the variables at the input nodes for determining the branches interconnected to the nodes to claim 1, which recites a judgement of how to select variables for the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 10 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 11,
Step 1 - “Is the claim to a process, machine, manufacture or composition of matter?”
Yes, the claim is directed towards a manufacture.
Step 2A, Prong 1 - “Is the claim directed to a law of nature, a natural phenomenon (product of nature) or an abstract idea?”:
The limitation of … determine a risk assessment decision for an operation using a set of decision trees based on a plurality of variables associated with a user, wherein the decision trees include pluralities of nodes interconnected by branches, the decision trees beginning with input nodes and ending with output nodes having decision results, wherein distinct decision results for the decision trees are determined based on the decision results at the output nodes, and wherein the variables have corresponding nodes in the decision trees; recites an evaluation of risk and a set of decision trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of assessing the data to determine at least one variable that does not have existing data in the accessed data; recites an evaluation of variables and their data within a dataset, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining the at least one variable that does not have existing data to be at least one deprecated variable for the set of decision trees; recites an evaluation of variables within a dataset, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of in response to determining the at least one deprecated variable, pruning at least one branch in the decision trees in the set of decision trees that include the at least one deprecated variable, wherein pruning the decision trees include pruning after an intermediate node based on the at least one deprecated variable, recites a judgement of what to prune within the trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of and wherein the intermediate node is replaced with an output node that provides a decision result based on a majority of decision results made at the intermediate node before the pruning; recites an evaluation of pruning the tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining distinct decision results at the output nodes recites an evaluation of the decision trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining a risk prediction based on a combination of the distinct decision results in the set of decision trees recites an evaluation of the result of the decision trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of and determining the risk assessment decision for the user based on the determined risk prediction for the user recites an evaluation of the result of the risk prediction, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Step 2A, Prong 2 - “Does the claim recite additional elements that integrate the judicial exception into a practical application?”:
The limitation of receiving a request to determine a risk assessment decision for an operation using a set of decision trees based on a plurality of variables associated with a user, recites the mere extra-solution activity of data gathering, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g).
The limitation of accessing data for the variables associated with the user recites the mere extra-solution activity of data gathering, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g).
The limitation of providing the data to the set of decision trees; recites the mere extra-solution activity of providing data to a decision tree for classification, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g).
Step 2B - “Does the claim recite additional elements that amount to significantly more than the judicial exception?”:
The limitation of receiving a request to determine a risk assessment decision for an operation using a set of decision trees based on a plurality of variables associated with a user, recites receiving data over a network, which is well-understood, routine, and conventional, MPEP 2106.05(d).II.i.
The limitation of accessing data for the variables associated with the user recites retrieving information in memory, which is well-understood, routine, and conventional, MPEP 2106.05(d).II.iv.
The limitation of providing the data to a set of decision trees; recites providing data to an ensemble of decision trees for classification, which is well-understood, routine, and conventional, (Ross “Why Do Tree Ensembles Work?” Pg. 1) “Ensembles of decision trees (e.g., the random forest and AdaBoost algorithms) are powerful and well-known methods of classification and regression”
Therefore, claim 11 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 12,
Claim 12 adds the additional limitation of wherein the data for the variables is accessed in response to receiving the request to claim 11, which recites a judgement of when to access the data, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 12 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 13,
Claim 12 adds the additional limitation of determining the at least one deprecated variable is deprecated based on the accessed data not including data for the at least one deprecated variable to claim 11, which recites a judgement of when a variable is deprecated, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Regarding claim 14,
Claim 14 adds the additional limitation of wherein the at least one decision tree is pruned after the intermediate node based on the intermediate node providing a decision result based on the at least one deprecated variable to claim 13, which recites a judgement of where to prune the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 14 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 15,
Claim 15 adds the additional limitations to claim 11:
receiving changes in information available for determining the risk assessment decision recites the mere extra-solution activity of data gathering, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g), and which recites receiving data over a network, which is well-understood, routine, and conventional, MPEP 2106.05(d).II.i.
deprecating at least one variable in the accessed data for the variables associated with the user, wherein the at least one variable is deprecated based on changes in information available for determining the risk assessment decision recites an evaluation of the variables, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
and pruning the decision tree after the intermediate node based on the intermediate node providing a decision result based on the at least one deprecated variable recites a judgement of where to prune the decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 15 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 16,
Step 1 - “Is the claim to a process, machine, manufacture or composition of matter?”
Yes, the claim is directed towards a process.
Step 2A, Prong 1 - “Is the claim directed to a law of nature, a natural phenomenon (product of nature) or an abstract idea?”:
The limitation of … determine a risk assessment decision using a set of decision trees for an operation associated with a user, wherein the decision trees include pluralities of nodes interconnected by branches, the decision trees beginning with input nodes and ending with output nodes having decision results, wherein distinct decision results for the decision trees are determined based on the decision results at the output nodes, recites an evaluation of risk and a decision tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of assessing, at the computer system, the dataset to determine at least one variable in the dataset that has data removed from the dataset; recites an evaluation of variables and their data within a dataset, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining the at least one variable in the dataset that has data removed from the dataset to be at least one deprecated variable for the decision trees; recites an evaluation of variables within a dataset, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of in response to determining the at least one deprecated variable in the dataset, pruning at least one branch in the decision trees in the set of decision trees that include the at least one deprecated variable, wherein pruning the decision trees include pruning after an intermediate node based on the at least one deprecated variable in the dataset, recites a judgement of what to prune within the trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of and wherein the intermediate node is replaced with an output node that provides a decision result based on a majority of decision results made at the intermediate node before the pruning; recites an evaluation of pruning the tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining a risk prediction based on a combination of the distinct decision results in the set of decision trees recites an evaluation of the result of the decision trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of and determining, by the computer system, the risk assessment decision for the user based on the determined risk prediction recites an evaluation of the result of the risk prediction, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Step 2A, Prong 2 - “Does the claim recite additional elements that integrate the judicial exception into a practical application?”:
The limitation of receiving, by a computer system, a request to determine a risk assessment decision for an operation associated with a user,…and wherein the request includes a dataset of variables associated with the user, the variables having corresponding nodes in the decision trees, and wherein the dataset of variables in the request has data for at least one variable removed from the dataset and being unavailable to the computer system recites the mere extra-solution activity of data gathering, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g).
The limitation of providing the dataset to a set of decision trees wherein the set of decision trees includes at least: a first decision tree having at least one branch pruned; and a second decision tree not having any pruned branches; recites the mere extra-solution activity of providing data to a set of decision trees for classification, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g).
Step 2B - “Does the claim recite additional elements that amount to significantly more than the judicial exception?”:
The limitation of receiving, by a computer system, a request to determine a risk assessment decision for an operation associated with a user,…and wherein the request includes a dataset of variables associated with the user, the variables having corresponding nodes in the decision trees, and wherein the dataset of variables in the request has data for at least one variable removed from the dataset and being unavailable to the computer system recites receiving data over a network, which is well-understood, routine, and conventional, MPEP 2106.05(d).II.i.
The limitation of providing the dataset to a set of decision trees wherein the set of decision trees includes at least: a first decision tree having at least one branch pruned; and a second decision tree not having any pruned branches; recites providing data to an ensemble of decision trees for classification, which is well-understood, routine, and conventional, (Ross “Why Do Tree Ensembles Work?” Pg. 1) “Ensembles of decision trees (e.g., the random forest and AdaBoost algorithms) are powerful and well-known methods of classification and regression”
Therefore, claim 16 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 17,
Claim 17 adds the additional limitation of wherein the set of decision trees includes: a third decision tree having two or more branches pruned after intermediate nodes that provide decision results based on the at least one deprecated variable, the nodes at ends of the pruned branches providing decision results based on majorities of previous decision results made at the intermediate nodes before the two or more branches have been pruned to claim 16, which recites a judgement of what the set of decision trees should include and of what to prune within the third tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 17 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 18,
Claim 18 adds the additional limitation of wherein the risk prediction is determined by averaging the distinct decision results in the set of decision trees to claim 16, which recites an evaluation of the distinct decision results of the decision trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 18 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 19,
Claim 19 adds the additional limitation of wherein the risk prediction is determined by determining a majority decision result from the distinct decision results in the set of decision trees to claim 16, which recites an evaluation of the distinct decision results of the decision trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 19 is found to be ineligible subject matter under 35 U.S.C. 101.
Regarding claim 20,
Claim 20 adds the additional limitations to claim 16:
The limitation of receiving, by the computer system, a second request to determine a second risk assessment decision for a second operation associated with a second user, wherein the request includes a second dataset of variables associated with the second user recites the mere extra-solution activity of data gathering, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g), and which recites receiving data over a network, which is well-understood, routine, and conventional, MPEP 2106.05(d).II.i.
The limitation of and wherein the second dataset of variables in the second request has data for a second deprecated variable removed from the second dataset, the second deprecated variable being different than the at least one deprecated variable recites a judgement on what to include within the dataset of variables, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of providing the dataset to the set of decision trees recites the mere extra-solution activity of providing data to a decision tree for classification, which does not integrate the exception into a practical application, MPEP 2106.05(d) and 2106.05(g), and which recites providing data to an ensemble of decision trees for classification, which is well-understood, routine, and conventional, (Ross “Why Do Tree Ensembles Work?” Pg. 1) “Ensembles of decision trees (e.g., the random forest and AdaBoost algorithms) are powerful and well-known methods of classification and regression”
The limitation of wherein the set of decision trees includes: a third decision tree having at least one branch pruned after an intermediate node that provides a decision result based on the second deprecated variable, the node at an end of the pruned branch providing a decision result based on a majority of previous decision results made at the intermediate node before the at least one branch has been pruned recites a judgement of what the set of decision trees should include and of what to prune within the third tree, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of determining a second risk prediction based on the combination of the distinct decision results in the set of decision trees recites an evaluation of the result of the decision trees, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
The limitation of and determining, by the computer system, the second risk assessment decision for the second user based on the second determined risk prediction recites an evaluation of the result of the risk prediction, which is a mental process, which is an abstract idea, regardless of if it’s performed on a generic computer.
Therefore, claim 20 is found to be ineligible subject matter under 35 U.S.C. 101.
Claim Rejections - 35 USC § 103
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 1-6 and 9 are rejected under 35 U.S.C. 103 as being unpatentable over Huang et al. (U.S. Patent Application Publication No. 2022/0035908), hereinafter Huang, in view of Vinayagasundaram et al. “Efficient Gaussian Decision Tree Method for Concept Drift Data Stream”, hereinafter Vinayagasundaram, further in view of Wang (U.S. Patent No. RE42,481), hereinafter Wang.
Regarding claim 1,
Huang teaches A method, comprising:
receiving, by a computer system, a request to determine a risk assessment decision for an operation associated with a user ((Huang [0020]) “the request module 132 may receive or detect a new IT change request. For example, the request module 132 may receive an indication of a new IT change request created by a user interacting with the GUI 145. In response to the new IT change request, the request module 132 may perform or initiate a risk classification of the change request”) using a decision tree, ((Huang Abstract) “in response to a receipt of the input data, perform a first risk classification of the IT change request using a decision tree model;”)
wherein the decision tree includes a plurality of nodes interconnected by branches, the decision tree beginning with one or more input nodes and ending with a plurality of output nodes having decision results, (Huang Fig. 3A shows a decision tree with a plurality of nodes, including a beginning input node (Success Rate > 90%), and several ending output nodes with decision results (L, M, or H))
PNG
media_image1.png
738
968
media_image1.png
Greyscale
and wherein the request includes a dataset of variables ((Huang [0010]) “A graphical user interface may be generated based on the risk classification, and may allow a user to evaluate risk impacts for various request features”) associated with the user, ((Huang [0020]) “the request module 132 may receive an indication of a new IT change request created by a user”, broadest reasonable interpretation of a dataset of variables associated with a user includes a set of request features for a user’s request) the variables having corresponding nodes in the decision tree; ((Huang [0020]) “the decision tree model may be traversed by comparing the request features of the received IT change request to the attribute tests represented by the non-leaf nodes of the decision tree model”)
providing the dataset to the decision tree; ((Huang [0020]) “the decision tree model may be traversed by comparing the request features of the received IT change request to the attribute tests represented by the non-leaf nodes of the decision tree model”)
determining distinct decision results at the output nodes; ((Huang [0019]) “a non-leaf node of the tree, may correspond to a binary test (e.g., ‘yes’ or ‘no,’ ‘true’ or ‘false,’ etc.)”, (Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category (e.g., high, medium, low), a number (e.g., 0.8, 0.5, etc.), or in any other form”, both non-leaf (intermediate) nodes and leaf nodes have distinct decision results)
determining a risk prediction based on a combination of the distinct decision results in the decision tree; ((Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category ( e.g., high, medium, low), a number (e.g., 0.8, 0.5, etc.), or in any other form”, Huang Fig. 3A shows the ultimate risk prediction is based on a combination of the distinct decision results throughout the path to the leaf node, broadest reasonable interpretation of a risk prediction includes a risk score, combining the distinct decision results is interpreted as either combining both decisions of non-leaf (intermediate) nodes and leaf nodes or combining the leaf nodes arrived at for each subtree of a decision tree, as shown in Fig. 10 of the instant application)
and determining, by the computer system, the risk assessment decision ((Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category (e.g., high, medium, low)”) for the user based on the determined risk prediction for the user ((Huang [0020]) “the request module 132 may receive or detect a new IT change request. For example, the request module 132 may receive an indication of a new IT change request created by a user interacting with the GUI 145. In response to the new IT change request, the request module 132 may perform or initiate a risk classification of the change request”, broadest reasonable interpretation of a risk assessment decision for a user includes a risk classification of a request made by a user)
Vinayagasundaram teaches the following further limitations that Huang does not teach:
in response to determining the at least one deprecated variable in the dataset, pruning at least one branch in the decision tree, wherein the decision tree is pruned after an intermediate node ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, replacing a node that splits into a branch with a leaf, i.e. an output node, corresponds to pruning a branch) based on the at least one deprecated variable in the dataset, ((Vinayagasundaram Pg. 1) “A data stream is an prearranged sequence of instances arriving at a rate that prevents them to permanently be accumulated them in memory…The key data stream characteristics entails the following constraints:…iii. The distribution that is generating the items will change over time. Thus, data from the past becomes irrelevant or harmful for the current samples…The vital point in constructing the decision tree is choosing the finest attribute to split the considered node”, instances with attributes that cannot be permanently stored and that have distributions that change over time correspond to deprecation of dataset variables)
and wherein the intermediate node is replaced with an output node that provides a decision result based on a majority of decision results made at the intermediate node before the pruning; ((Vinayagasundaram Pg. 3) “Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics”, a leaf is an output node, the statistically most common class at a node corresponds to a decision result based on a majority of decision results)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang and Vinayagasundaram by taking the decision tree for risk prediction taught by Huang, and adding the use of pruning of branches of the decision tree based on variable deprecation due to concept drift in a data stream taught by Vinayagasundaram, as Vinayagasundaram teaches: (Vinayagasundaram Pg. 3) “One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics. This policy assures that GDT will always remain as current as possible with respect to the underlying stream generating distribution”. Such a combination would be obvious.
Wang teaches the following further limitations that Huang does not teach and more explicitly than Vinayagasundaram teaches:
assessing, at the computer system, the dataset to determine at least one variable in the dataset that no longer has data existing in the dataset; ((Wang Col. 2, lines 23-26) “When a data set comes in, it first goes through a data preprocessing step in which the validity of the data in the data set is checked and cases or parameters with missing data are eliminated”)
determining, at the computer system, the at least one variable in the dataset that no longer has data existing in the dataset to be at least one deprecated variable ((Wang Col. 4, lines 25-30) “It is very common for a yield data set to have missing values. The data pre-processing step removes the cases or variables having missing values. In particular, the preprocessing first may remove all predictor variables that are ‘bad’. By ‘bad’, it is understood that either a variable has too much missing data”) for the decision tree ((Wang Col. 4, lines 61-67) “Thus, once the preprocessing has been completed, the data set 54 contains no missing values so that the statistical analysis may produce better results. Now, the model building step in accordance with the invention will be described in more detail. The yield management system uses a decision tree-based method”)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, and Wang by taking the decision tree for risk prediction with branches pruned based on variable deprecation, jointly taught by Huang and Vinayagasundaram, and having the variable deprecation consist of data for a variable no longer existing in the dataset, taught by Wang, as it is well known in the art that variables/features used for machine learning model inference that have an excessive amount of missing data become counterproductive when used to make predictions, as most data instances can only be classified without the variable/feature, and so detecting the issue and deprecating the variable increases the robustness and accuracy of the model in practice. Such a combination would be obvious.
Regarding claim 2,
Huang, Vinayagasundaram, and Wang jointly teach The method of claim 1,
Huang further teaches:
wherein the dataset of variables in the request has data for the at least one [deprecated] variable removed from the dataset ((Huang [0033]) “In other examples, the controls 335 may be used to manually add or remove request features from the decision tree model”, Vinayagasundaram teaches variable deprecation) after generation of the decision tree ((Huang [0033]) “Further, adding and/or removing request features may cause the decision tree model to be automatically re-trained according to the new set of request features”, re-training the decision tree after removing request features corresponds to the request features being removed after initial training, and thus generation, of the decision tree)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, and Wang for the parent claim of claim 2, claim 1. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 3,
Huang, Vinayagasundaram, and Wang jointly teach The method of claim 1,
Vinayagasundaram further teaches:
wherein the intermediate node for the pruning is a node providing a decision result based on the at least one deprecated variable ((Vinayagasundaram Pg. 3) “AGDT periodically scans the GDT and all the alternate trees glancing for internal node's statistics to indicate some new attribute that would be a best attribute than the chosen split attribute…In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics. This policy assures that GDT will always remain as current as possible with respect to the underlying stream generating distribution”, an internal node that splits on an attribute, where the attribute is no longer the best attribute to split on, resulting in the node being pruned corresponds to pruning an intermediate node that provides a decision based on a deprecated variable)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, and Wang for the parent claim of claim 3, claim 1. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 4,
Huang, Vinayagasundaram, and Wang jointly teach The method of claim 1, further comprising:
Vinayagasundaram further teaches:
deprecating the at least one variable in the dataset of variables in the request ((Vinayagasundaram Pg. 3) “AGDT periodically scans the GDT and all the alternate trees glancing for internal node's statistics to indicate some new attribute that would be a best attribute than the chosen split attribute”, broadest reasonable interpretation of deprecating a variable corresponds to choosing to not use it for a decision it was previously used for)
wherein the at least one variable is deprecated based on changes in information available for determining the [risk assessment] decision ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics. This policy assures that GDT will always remain as current as possible with respect to the underlying stream generating distribution”, change in concept corresponds to changes in information, Huang teaches risk assessment decisions)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, and Wang for the parent claim of claim 4, claim 1. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 5,
Huang, Vinayagasundaram, and Wang jointly teach The method of claim 1,
Vinayagasundaram further teaches:
wherein the decision tree includes a plurality of branches with intermediate nodes providing decision results based on the at least one deprecated variable ((Vinayagasundaram Pg. 3) “If the concept is shifting, some split samples will no longer appear to be best because new data may provide more gain than previous one. In such circumstances, AGDT creates an alternative sub-tree to find best attribute at root…AGDT periodically scans the GDT and all the alternate trees glancing for internal node's statistics to indicate some new attribute that would be a best attribute than the chosen split attribute”, the decision tree having several alterative sub-trees with splits at internal nodes based on attributes that are losing relevant corresponds to a decision tree with a plurality of branches with intermediate nodes that decide based on deprecated variables)
the method further comprising: pruning the plurality of branches in the decision tree ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, checking every node split and replacing nodes that splits into branches with leaf nodes, i.e. output nodes, corresponds to pruning the plurality of branches)
wherein the decision trees are pruned after the intermediate nodes providing decision results based on the at least one deprecated variable ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, replacing a node that splits into a branch with a leaf, i.e. an output node, corresponds to pruning a branch)
and wherein the intermediate nodes are replaced with output nodes that provide decision results based on majorities of decision results made at the intermediate nodes before the pruning ((Vinayagasundaram Pg. 3) “Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics”, a leaf is an output node, the statistically most common class at a node corresponds to a decision result based on a majority of decision results)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, and Wang for the parent claim of claim 5, claim 1. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 9,
Huang, Vinayagasundaram, and Wang jointly teach The method of claim 1,
Vinayagasundaram further teaches:
further comprising pruning, after a set of decision results, one or more branches in the decision tree that lack prediction power in the set of decision results ((Vinayagasundaram Pg. 3) “AGDT periodically scans the GDT and all the alternate trees glancing for internal node's statistics to indicate some new attribute that would be a best attribute than the chosen split attribute…In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics. This policy assures that GDT will always remain as current as possible with respect to the underlying stream generating distribution”, replacing internal nodes that split on an attribute, where the attribute is no longer the best attribute to split on, with leaf nodes corresponds to pruning branches of the decision tree that lack prediction power)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, and Wang for the parent claim of claim 9, claim 1. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Claims 7, 8, and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Huang, in view of Vinayagasundaram, further in view of Wang, further in view of Tsou et al. (U.S. Patent No. 11,315,045), hereinafter Tsou.
Regarding claim 7,
Huang, Vinayagasundaram, and Wang jointly teach The method of claim 1,
Tsou teaches the following further limitation that neither Huang, nor Vinayagasundaram, nor Wang teach:
wherein the [risk] prediction is determined by averaging the distinct decision results in the decision tree ((Tsou Col. 15, lines 31-39) “A random forest may be considered an ensemble model based on multiple decision trees...an ensemble may include a voting algorithm, in which each model provides a vote and a majority, plurality, mean, or median vote may be adopted as the aggregate result of the collection of models”, mean voting would be averaging the decision tree results, determining a majority decision from decision results in the decision tree is interpreted as combining the leaf nodes arrived at for each subtree of a decision tree, as shown in Fig. 10 of the instant application, in a manner similar to combining the leaf nodes arrived at in each decision tree of an ensemble, Huang teaches risk prediction)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, Wang, and Tsou by taking the decision tree for risk prediction, where nodes related to depreciated variables are pruned, jointly taught by Huang, Vinayagasundaram, and Wang, and using it within an ensemble of decision subtrees where the final result is determined by averaging the predictions of each tree, as taught by Tsou, as doing so is a well-known technique within the art for combining the simplicity and efficiency of simple machine learning models with the accuracy of more complex machine learning models, and producing a final result from such an ensemble. Such a combination would be obvious.
Regarding claim 8,
Huang, Vinayagasundaram, and Wang jointly teach The method of claim 1,
Tsou teaches the following further limitation that neither Huang, nor Vinayagasundaram, nor Wang explicitly teach:
wherein the [risk] prediction is determined by determining a majority decision result from the distinct decision results in the decision tree ((Tsou Col. 15, lines 31-39) “A random forest may be considered an ensemble model based on multiple decision trees...an ensemble may include a voting algorithm, in which each model provides a vote and a majority, plurality, mean, or median vote may be adopted as the aggregate result of the collection of models”, majority voting would be determining a majority decision result from the decision tree results, determining a majority decision from decision results in the decision tree is interpreted as combining the leaf nodes arrived at for each subtree of a decision tree, as shown in Fig. 10 of the instant application, in a manner similar to combining the leaf nodes arrived at in each decision tree of an ensemble, Huang teaches risk prediction)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, Wang, and Tsou by taking the decision tree for risk prediction, where nodes related to depreciated variables are pruned, jointly taught by Huang, Vinayagasundaram, and Wang, and using it within an ensemble of decision subtrees where the final result is determined by the majority vote of the predictions of each tree, as taught by Tsou, as doing so is a well-known technique within the art for combining the simplicity and efficiency of simple machine learning models with the accuracy of more complex machine learning models, and producing a final result from such an ensemble. Such a combination would be obvious.
Regarding claim 10,
Huang, Vinayagasundaram, and Wang jointly teach The method of claim 1,
Tsou teaches the following further limitation that neither Huang, nor Vinayagasundaram, nor Wang explicitly teaches:
wherein the decision tree includes a random selection of the variables at the input nodes for determining the branches interconnected to the nodes ((Tsou Col. 15, lines 49-52) “Training a decision tree can involve randomly selecting features to assist in reducing the amount of generated 50 positives in training time. Indeed, through sub-featuring, the size of the training data domain may be reduced”, broadest reasonable interpretation of random selection of variables at input nodes includes randomly choosing a subset of the variables to use in the tree, including immediately after the input nodes)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, Wang, and Tsou by taking the decision tree for risk prediction, where nodes related to depreciated variables are pruned, as jointly taught by Huang, Vinayagasundaram, and Wang, and randomly applying the variables at the nodes and branches of the tree, as taught by Tsou, as Tsou teaches: (Tsou Col. 15, lines 42-48) “each decision tree may be trained by…sub-featuring (e.g., random selecting training features), to ensure that sufficient variety exists in the various decision trees implementing a proposed random forest model (e.g., to improve the results derived through the ensemble).” Such a combination would be obvious.
Claims 11-15 are rejected under 35 U.S.C. 103 as being unpatentable over Huang, in view of Vinayagasundaram, further in view of Wang, further in view of Barddal et al. “A Survey on Feature Drift Adaptation”, hereinafter Barddal.
Regarding claim 11,
Huang teaches A non-transitory computer-readable medium having instructions stored thereon that are executable by a computing device to perform operations, comprising: ((Huang [0038]) “The process 400 may be implemented in hardware or machine-readable instructions (e.g., software and/or firmware). The machine-readable instructions are stored in a non-transitory computer readable medium, such as an optical, semiconductor, or magnetic storage device”)
receiving a request to determine a risk assessment decision for an operation using [a set of] decision trees based on a plurality of variables associated with a user, ((Huang [0020]) “the request module 132 may receive or detect a new IT change request. For example, the request module 132 may receive an indication of a new IT change request created by a user interacting with the GUI 145. In response to the new IT change request, the request module 132 may perform or initiate a risk classification of the change request using the decision tree model”, Huang teaches one but not a set of decision trees)
wherein the decision trees include pluralities of nodes interconnected by branches, the decision trees beginning with input nodes and ending with output nodes having decision results, (Huang Fig. 3A shows a decision tree with a plurality of nodes, including a beginning input node (Success Rate > 90%), and several ending output nodes with decision results (L, M, or H))
wherein distinct decision results for the decision trees are determined based on the decision results at the output nodes, ((Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category ( e.g., high, medium, low), a number (e.g., 0.8, 0.5, etc.), or in any other form”)
and wherein the variables have corresponding nodes in the decision trees; ((Huang [0020]) “the decision tree model may be traversed by comparing the request features of the received IT change request to the attribute tests represented by the non-leaf nodes of the decision tree model”)
accessing data for the variables ((Huang [0026]) “the decision tree model may be traversed by comparing the request features of the received IT change request to the attribute tests represented by the non-leaf nodes of the decision tree”, comparing the request features would require accessing data for them) associated with the user ((Huang [0020]) “the request module 132 may receive an indication of a new IT change request created by a user”, broadest reasonable interpretation of variables associated with a user includes a set of request features for a user’s request)
determining distinct decision results at the output nodes; ((Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category ( e.g., high, medium, low), a number (e.g., 0.8, 0.5, etc.), or in any other form”)
and determining the risk assessment decision ((Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category ( e.g., high, medium, low”) for the user based on the determined risk prediction for the user ((Huang [0020]) “the request module 132 may receive or detect a new IT change request. For example, the request module 132 may receive an indication of a new IT change request created by a user interacting with the GUI 145. In response to the new IT change request, the request module 132 may perform or initiate a risk classification of the change request”, broadest reasonable interpretation of a risk assessment decision for a user includes a risk classification of a request made by a user)
Vinayagasundaram teaches the following further limitations that Huang does not teach:
in response to determining the at least one deprecated variable, pruning at least one branch in the decision trees in the set of decision trees that include the at least one deprecated variable, wherein pruning the decision trees include pruning after an intermediate node ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, replacing a node that splits into a branch with a leaf, i.e. an output node, corresponds to pruning a branch) based on the at least one deprecated variable, ((Vinayagasundaram Pg. 1) “A data stream is an prearranged sequence of instances arriving at a rate that prevents them to permanently be accumulated them in memory…The key data stream characteristics entails the following constraints:…iii. The distribution that is generating the items will change over time. Thus, data from the past becomes irrelevant or harmful for the current samples…The vital point in constructing the decision tree is choosing the finest attribute to split the considered node”, instances with attributes that cannot be permanently stored and that have distributions that change over time correspond to deprecation of variables)
and wherein the intermediate node is replaced with an output node that provides a decision result based on a majority of decision results made at the intermediate node before the pruning; ((Vinayagasundaram Pg. 3) “Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics”, a leaf is an output node, the statistically most common class at a node corresponds to a decision result based on a majority of decision results)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang and Vinayagasundaram by taking the decision tree for risk prediction taught by Huang, and adding the use of pruning of branches of the decision tree based on variable deprecation due to concept drift in a data stream taught by Vinayagasundaram, as Vinayagasundaram teaches: (Vinayagasundaram Pg. 3) “One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics. This policy assures that GDT will always remain as current as possible with respect to the underlying stream generating distribution”. Such a combination would be obvious.
Wang teaches the following further limitations that Huang does not teach and more explicitly than Vinayagasundaram teaches:
assessing the data to determine at least one variable that does not have existing data in the accessed data; ((Wang Col. 2, lines 23-26) “When a data set comes in, it first goes through a data preprocessing step in which the validity of the data in the data set is checked and cases or parameters with missing data are eliminated”)
determining the at least one variable that does not have existing data to be at least one deprecated variable ((Wang Col. 4, lines 25-30) “It is very common for a yield data set to have missing values. The data pre-processing step removes the cases or variables having missing values. In particular, the preprocessing first may remove all predictor variables that are ‘bad’. By ‘bad’, it is understood that either a variable has too much missing data”) for the [set of] decision trees; ((Wang Col. 4, lines 61-67) “Thus, once the preprocessing has been completed, the data set 54 contains no missing values so that the statistical analysis may produce better results. Now, the model building step in accordance with the invention will be described in more detail. The yield management system uses a decision tree-based method”, Wang teaches one but not a set of decision trees)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, and Wang by taking the decision tree for risk prediction with branches pruned based on variable deprecation, jointly taught by Huang and Vinayagasundaram, and having the variable deprecation consist of data for a variable no longer existing in the dataset, taught by Wang, as it is well known in the art that variables/features used for machine learning model inference that have an excessive amount of missing data become counterproductive when used to make predictions, as most data instances can only be classified without the variable/feature, and so detecting the issue and deprecating the variable increases the robustness and accuracy of the model in practice. Such a combination would be obvious.
Barddal teaches the following further limitations that neither Huang, nor Vinayagasundaram, nor Wang teach:
providing the data to the set of decision trees; ((Barddal Pg. 4) “Since a random forest is an ensemble classifier, the classification of each new instance is the fusion of the votes of the containing trees”, an instance is data, and a random forest includes a set of decision trees)
determining a [risk] prediction based on a combination of the distinct decision results in the set of decision trees; ((Barddal Pg. 4) “Since a random forest is an ensemble classifier, the classification of each new instance is the fusion of the votes of the containing trees”, combining the distinct decision results is interpreted as either combining only the final decision for each tree in the ensemble or combining the leaf nodes arrived at for each subtree of each decision tree within the ensemble, as shown in Fig. 10 of the instant application, Huang teaches risk prediction)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, Wang, and Barddal by taking the decision tree for risk prediction, which is pruned based on variable deprecation, jointly taught by Huang, Vinayagasundaram, and Wang, and using multiple of them within an ensemble, as taught by Barddal, as using an ensemble of several simple machine learning models, such as decision trees, and combining each of their results for a given input to make a final prediction is a well-known technique within the art for combining the simplicity and efficiency of simple machine learning models with the accuracy of more complex machine learning models. Such a combination would be obvious.
Regarding claim 12,
Huang, Vinayagasundaram, Wang, and Barddal jointly teach The non-transitory computer-readable medium of claim 11,
Huang further teaches:
wherein the data for the variables is accessed in response to receiving the request ((Huang [0026]) “In response to the new IT change request, the request module 132 may perform or initiate a risk classification of the change request using the decision tree model...the decision tree model may be traversed by comparing the request features of the received IT change request to the attribute tests represented by the non-leaf nodes of the decision tree”, comparing the variables/features to the attribute tests requires accessing the data for the variables/features)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, Wang, and Barddal for the parent claim of claim 12, claim 11. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 13,
Huang, Vinayagasundaram, Wang, and Barddal jointly teach The non-transitory computer-readable medium of claim 11, further comprising:
Wang further teaches:
determining the at least one deprecated variable is deprecated based on the accessed data not including data for the at least one deprecated variable ((Wang Col. 4, lines 25-30) “It is very common for a yield data set to have missing values. The data pre-processing step removes the cases or variables having missing values. In particular, the preprocessing first may remove all predictor variables that are ‘bad’. By ‘bad’, it is understood that either a variable has too much missing data”)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, Wang, and Barddal for the parent claim of claim 13, claim 11. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 14,
Huang, Vinayagasundaram, Wang, and Barddal jointly teach The non-transitory computer-readable medium of claim 13,
Vinayagasundaram further teaches:
wherein the at least one decision tree is pruned after the intermediate node ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, replacing an internal node that splits into a branch with a leaf, i.e. an output node, corresponds to pruning the decision tree after the intermediate node)
based on the intermediate node providing a decision result based on the at least one deprecated variable ((Vinayagasundaram Pg. 3) “AGDT periodically scans the GDT and all the alternate trees glancing for internal node's statistics to indicate some new attribute that would be a best attribute than the chosen split attribute”)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, Wang, and Barddal for the parent claim of claim 14, claim 13. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Regarding claim 15,
Huang, Vinayagasundaram, Wang, and Barddal jointly teach The non-transitory computer-readable medium of claim 11, further comprising:
Barddal further teaches:
receiving changes in information available for determining the [risk assessment] decision ((Barddal Pg. 2) “Due to the inherent temporal aspect of data streams, their underlying data distribution is expected to dynamically change over time, implying in changes in the concept to be learned, phenomenon named concept drift”, (Barddal Pg. 2) “Feature drifts occur whenever the relevance of a feature Di...grows or shrinks for incoming instances”, Huang teaches risk assessment decisions)
deprecating at least one variable in the accessed data for the variables [associated with the user] ((Barddal Abstract) “In this work, we focus on one specific kind of concept drift that has not been extensively addressed in the literature, namely feature drift. A feature drift happens when changes occur in the set of features, such that a subset of features become, or cease to be, relevant to the learning problem”, broadest reasonable interpretation of deprecation of a variable/feature includes the variable/feature no longer being relevant, Huang teaches variables/features associated with users)
wherein the at least one variable is deprecated based on changes in information available for determining the [risk assessment] decision ((Barddal Pg. 2) “Due to the inherent temporal aspect of data streams, their underlying data distribution is expected to dynamically change over time, implying in changes in the concept to be learned, phenomenon named concept drift”, (Barddal Pg. 2) “Feature drifts occur whenever the relevance of a feature Di...grows or shrinks for incoming instances”, Huang teaches risk assessment decisions)
Vinayagasundaram further teaches:
and pruning the decision tree after the intermediate node based on the intermediate node providing a decision result based on the at least one deprecated variable ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, replacing a node that splits into a branch with a leaf, i.e. an output node, corresponds to pruning a decision tree after the intermediate node)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, Wang, and Barddal for the parent claim of claim 15, claim 11. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Claims 16 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Huang, in view of Vinayagasundaram, further in view of Barddal, further in view of Gama et al. “Learning Decision Trees from Dynamic Data Streams”, hereinafter Gama.
Regarding claim 16,
Huang teaches A method, comprising:
receiving, by a computer system, a request to determine a risk assessment decision using [a set of] decision trees for an operation associated with a user, ((Huang [0020]) “the request module 132 may receive or detect a new IT change request. For example, the request module 132 may receive an indication of a new IT change request created by a user interacting with the GUI 145. In response to the new IT change request, the request module 132 may perform or initiate a risk classification of the change request using the decision tree model”, Huang teaches one but not a set of decision trees)
wherein the decision trees include pluralities of nodes interconnected by branches, the decision trees beginning with input nodes and ending with output nodes having decision results, (Huang Fig. 3A shows a decision tree with a plurality of nodes, including a beginning input node (Success Rate > 90%), and several ending output nodes with decision results (L, M, or H))
wherein distinct decision results for the decision trees are determined based on the decision results at the output nodes, ((Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category ( e.g., high, medium, low), a number (e.g., 0.8, 0.5, etc.), or in any other form”)
and wherein the request includes a dataset of variables ((Huang [0010]) “A graphical user interface may be generated based on the risk classification, and may allow a user to evaluate risk impacts for various request features”) associated with the user, ((Huang [0020]) “the request module 132 may receive an indication of a new IT change request created by a user”, broadest reasonable interpretation of a dataset of variables associated with a user includes a set of request features for a user’s request) the variables having corresponding nodes in the decision trees, ((Huang [0020]) “the decision tree model may be traversed by comparing the request features of the received IT change request to the attribute tests represented by the non-leaf nodes of the decision tree model”)
and wherein the dataset of variables in the request has data for at least one variable removed from the dataset [and unavailable to the computer system]; ((Huang [0033]) “In other examples, the controls 335 may be used to manually add or remove request features from the decision tree model”, Huang does not teach variable data removal causing the data to be unavailable to the computer system)
and determining, by the computer system, the risk assessment decision for the user (Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category ( e.g., high, medium, low”) based on the determined risk prediction ((Huang [0020]) “the request module 132 may receive or detect a new IT change request. For example, the request module 132 may receive an indication of a new IT change request created by a user interacting with the GUI 145. In response to the new IT change request, the request module 132 may perform or initiate a risk classification of the change request”, broadest reasonable interpretation of a risk assessment decision for a user includes a risk classification of a request made by a user)
Vinayagasundaram teaches the following further limitations that Huang does not teach:
… data for at least one variable removed from the dataset and unavailable to the computer system; ((Vinayagasundaram Pg. 1) “A data stream is an prearranged sequence of instances arriving at a rate that prevents them to permanently be accumulated them in memory…The key data stream characteristics entails the following constraints:…iii. The distribution that is generating the items will change over time. Thus, data from the past becomes irrelevant or harmful for the current sample”, instances with attributes that cannot be permanently stored and that have distributions that change over time corresponds to data of dataset variables that is unavailable to the computer system)
assessing, at the computer system, the dataset to determine at least one variable in the dataset that has data removed from the dataset; ((Vinayagasundaram Pg. 1) “A data stream is an prearranged sequence of instances arriving at a rate that prevents them to permanently be accumulated them in memory…The key data stream characteristics entails the following constraints:…iii. The distribution that is generating the items will change over time. Thus, data from the past becomes irrelevant or harmful for the current samples…The vital point in constructing the decision tree is choosing the finest attribute to split the considered node”, instances with attributes that cannot be permanently stored and that have distributions that change over time correspond to removal of data from dataset variables)
determining the at least one variable in the dataset that has data removed from the dataset to be at least one deprecated variable for the decision trees; ((Vinayagasundaram Pg. 3) “If the concept is shifting, some split samples will no longer appear to be best because new data may provide more gain than previous one. In such circumstances, AGDT creates an alternative sub-tree to find best attribute at root…AGDT periodically scans the GDT and all the alternate trees glancing for internal node's statistics to indicate some new attribute that would be a best attribute than the chosen split attribute”, variables with data that has been removed no longer being a best attribute for a split in a decision tree corresponds to determination of the variable as deprecated)
in response to determining the at least one deprecated variable in the dataset, pruning at least one branch in the decision trees in the set of decision trees that include the at least one deprecated variable, wherein pruning the decision trees include pruning after an intermediate node ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, replacing a node that splits into a branch with a leaf, i.e. an output node, corresponds to pruning a branch) based on the at least one deprecated variable in the dataset, ((Vinayagasundaram Pg. 1) “A data stream is an prearranged sequence of instances arriving at a rate that prevents them to permanently be accumulated them in memory…The key data stream characteristics entails the following constraints:…iii. The distribution that is generating the items will change over time. Thus, data from the past becomes irrelevant or harmful for the current samples…The vital point in constructing the decision tree is choosing the finest attribute to split the considered node”, instances with attributes that cannot be permanently stored and that have distributions that change over time correspond to deprecation of dataset variables)
and wherein the intermediate node is replaced with an output node that provides a decision result based on a majority of decision results made at the intermediate node before the pruning; ((Vinayagasundaram Pg. 3) “Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics”, a leaf is an output node, the statistically most common class at a node corresponds to a decision result based on a majority of decision results)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang and Vinayagasundaram by taking the decision tree for risk prediction taught by Huang, and adding the use of pruning of branches of the decision tree based on variable deprecation due to concept drift in a data stream taught by Vinayagasundaram, as Vinayagasundaram teaches: (Vinayagasundaram Pg. 3) “One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics. This policy assures that GDT will always remain as current as possible with respect to the underlying stream generating distribution”. Such a combination would be obvious.
Barddal teaches the following further limitations that neither Huang nor Vinayagasundaram teach:
providing the dataset to a set of decision trees, ((Barddal Pg. 4) “Since a random forest is an ensemble classifier, the classification of each new instance is the fusion of the votes of the containing trees”, an instance is data, and a random forest includes a set of decision trees)
determining a [risk] prediction based on a combination of the distinct decision results in the set of decision trees; ((Barddal Pg. 4) “Since a random forest is an ensemble classifier, the classification of each new instance is the fusion of the votes of the containing trees”, combining the distinct decision results is interpreted as either combining only the final decision for each tree in the ensemble or combining the leaf nodes arrived at for each subtree of each decision tree within the ensemble, as shown in Fig. 10 of the instant application, Huang teaches risk prediction)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, and Barddal by taking the decision tree for risk prediction, which is pruned based on variable deprecation, jointly taught by Huang and Vinayagasundaram, and using multiple of them within an ensemble, as taught by Barddal, as using an ensemble of several simple machine learning models, such as decision trees, and combining each of their results for a given input to make a final prediction is a well-known technique within the art for combining the simplicity and efficiency of simple machine learning models with the accuracy of more complex machine learning models. Such a combination would be obvious.
Gama teaches the following further limitations that neither Huang nor Vinayagasundaram teach and more explicitly than Barddal teaches:
wherein the set of decision trees includes at least: a first decision tree having at least one branch pruned; and a second decision tree not having any pruned branches; ((Gama Abstract) “This paper presents a system for induction of forest of functional trees from data streams able to detect concept drift. The Ultra Fast Forest of Trees (UFFT) is an incremental algorithm, that works online…Decision nodes and leaves contain naive-Bayes classifiers…naive-Bayes in inner nodes can be used as multivariate splitting-tests if chosen by the splitting criteria, and used to detect drift in the distribution of the examples that traverse the node. When a drift is detected, all the sub-tree rooted at that node will be pruned”, a sub-tree is a branch, a forest of decision trees that prunes trees with drift detected includes trees that do not have drift detected and thus are not pruned)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, Barddal, and Gama by taking the ensemble of decision trees for risk prediction, which is pruned based on variable deprecation, jointly taught by Huang, Vinayagasundaram, and Barddal, and having the ensemble include both trees that are pruned and trees that are not pruned, as taught by Gama, as doing so allows for optimal combination of trees in the ensemble that are suited for classifying different subsets of features of inputs, as is common in the art, by only pruning trees that include a feature that should not be used for predictions any longer, but not wasting computation on trees that do not include the feature and do not require pruning. Such a combination would be obvious.
Regarding claim 17,
Huang, Vinayagasundaram, Barddal, and Gama jointly teach The method of claim 16, wherein the set of decision trees includes:
Vinayagasundaram further teaches:
a [third] decision tree having two or more branches pruned after intermediate nodes that provide decision results based on the at least one deprecated variable ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, replacing a node that splits into a branch with a leaf, i.e. an output node, corresponds to pruning a branch, pruning more than one branch could occur, Barddal teaches an ensemble of decision trees which could include at least three trees)
the nodes at ends of the pruned branches providing decision results based on majorities of previous decision results made at the intermediate nodes before the two or more branches have been pruned ((Vinayagasundaram Pg. 3) “Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics”, a leaf is an output node, the statistically most common class at a node corresponds to a decision result based on a majority of decision results)
At the time of filing, one of ordinary skill in the art would have motivation to combine the method jointly taught by Huang, Vinayagasundaram, Barddal, and Gama for the parent claim of claim 17, claim 16. No new embodiments are introduced, so the reason to combine is the same as for the parent claim.
Claims 18 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Huang, in view of Vinayagasundaram, further in view of Barddal, further in view of Gama, further in view of Tsou.
Regarding claim 18,
Huang, Vinayagasundaram, Barddal, and Gama jointly teach The method of claim 16,
Tsou teaches the following further limitation that neither Huang, nor Vinayagasundaram, nor Barddal, nor Gama teaches:
wherein the [risk] prediction is determined by averaging the distinct decision results in the set of decision trees ((Tsou Col. 15, lines 31-39) “A random forest may be considered an ensemble model based on multiple decision trees...an ensemble may include a voting algorithm, in which each model provides a vote and a majority, plurality, mean, or median vote may be adopted as the aggregate result of the collection of models”, mean voting would be averaging the decision tree results, Huang teaches risk prediction)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, Barddal, Gama, and Tsou by taking the ensemble of decision tree for risk prediction, where nodes related to depreciated variables are pruned, jointly taught by Huang, Vinayagasundaram, Barddal, and Gama, and having the final result determined by averaging the predictions of each tree, as taught by Tsou, as doing so is a well-known technique within the art for producing a final result from an ensemble of trees, which is useful for fine-grained numerical data in which different trees in the ensemble getting precisely the same results is unlikely. Such a combination would be obvious.
Regarding claim 19,
Huang, Vinayagasundaram, Barddal, and Gama jointly teach The method of claim 16,
Tsou teaches the following further limitation that neither Huang, nor Vinayagasundaram, nor Barddal, nor Gama explicitly teaches:
wherein the [risk] prediction is determined by determining a majority decision result from the distinct decision results in the set of decision trees ((Tsou Col. 15, lines 31-39) “A random forest may be considered an ensemble model based on multiple decision trees...an ensemble may include a voting algorithm, in which each model provides a vote and a majority, plurality, mean, or median vote may be adopted as the aggregate result of the collection of models”, majority voting would be determining a majority decision result from the decision tree results, Huang teaches risk prediction)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, Barddal, Gama, and Tsou by taking the ensemble of decision tree for risk prediction, where nodes related to depreciated variables are pruned, as jointly taught by Huang, Vinayagasundaram, Barddal, and Gama, and having the final result determined by majority vote of the outputs of each tree, as taught by Tsou, as doing so is a well-known technique within the art for producing a final result from an ensemble of trees, which is useful when handling discrete data for which averaging would not yield a coherent result. Such a combination would be obvious.
Claims 20 is rejected under 35 U.S.C. 103 as being unpatentable over Huang, further in view of Vinayagasundaram, further in view of Barddal, further in view of Gama, further in view of Mislove (International Patent Application Publication No. 2013/138514), hereinafter Mislove.
Regarding claim 20,
Huang, Vinayagasundaram, Barddal, and Gama jointly teach The method of claim 16, further comprising:
Huang further teaches:
wherein the request includes a [second] dataset of variables ((Huang [0010]) “A graphical user interface may be generated based on the risk classification, and may allow a user to evaluate risk impacts for various request features”) associated with the [second] user ((Huang [0020]) “the request module 132 may receive an indication of a new IT change request created by a user”, broadest reasonable interpretation of a dataset of variables associated with a user includes a set of request features for a user’s request, a second user is taught by Mislove)
and wherein the [second] dataset of variables in the [second] request has data for a [second] [deprecated] variable removed from the [second] dataset ((Huang [0033]) “In other examples, the controls 335 may be used to manually add or remove request features from the decision tree model”, Vinayagasundaram and Barddal teach variable deprecation, a second request is taught by Mislove)
and determining, by the computer system, the [second] risk assessment decision ((Huang [0020]) “the destination node (i.e., the leaf node that is reached by traversing the decision tree model) may indicate a risk score for the received IT change request. The risk score may be represented as a category ( e.g., high, medium, low”) for the [second] user based on the [second] determined risk prediction ((Huang [0020]) “the request module 132 may receive or detect a new IT change request. For example, the request module 132 may receive an indication of a new IT change request created by a user interacting with the GUI 145. In response to the new IT change request, the request module 132 may perform or initiate a risk classification of the change request”, broadest reasonable interpretation of a risk assessment decision for a user includes a risk classification of a request made by a user, a second user is taught by Mislove)
Vinayagasundaram further teaches:
a [third] decision tree having at least one branch pruned after an intermediate node that provides a decision result based on the second deprecated variable, ((Vinayagasundaram Pg. 3) “In the internal nodes of GDT validity at each node split is checked from time to time…Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf”, replacing a node that splits into a branch with a leaf, i.e. an output node, corresponds to pruning a branch, Barddal teaches an ensemble of decision trees which could include at least three trees)
the node at an end of the pruned branch providing a decision result based on a majority of previous decision results made at the intermediate node before the at least one branch has been pruned; ((Vinayagasundaram Pg. 3) “Let us consider a node 'n' where a change in concept is detected. One solution to handle this concept change is to replace this node with a leaf predicting the most common class in node's sufficient statistics”, a leaf is an output node, the statistically most common class at a node corresponds to a decision result based on a majority of decision results)
Barddal further teaches:
the second deprecated variable being different than the at least one deprecated variable ((Barddal Abstract) “A feature drift happens when changes occur in the set of features, such that a subset of features become, or cease to be, relevant to the learning problem”, a subset of features becoming irrelevant can include more than one feature becoming irrelevant)
providing the dataset to the set of decision trees, wherein the set of decision trees includes: a third decision tree ((Barddal Pg. 4) “Since a random forest is an ensemble classifier, the classification of each new instance is the fusion of the votes of the containing trees”, an instance is part of a dataset, and a random forest includes a set of decision trees, of which there can be at least three)
determining a second [risk] prediction based on the combination of the distinct decision results in the set of decision trees ((Barddal Pg. 4) “Since a random forest is an ensemble classifier, the classification of each new instance is the fusion of the votes of the containing trees”, there can be more than one instance to be classified, Huang teaches risk prediction)
Mislove teaches the following further limitation that neither Huang, nor Vinayagasundaram, nor Barddal, nor Gama explicitly teaches:
receiving, by the computer system, a second request to determine a second risk assessment decision for a second operation associated with a second user ((Mislove [0010]) “the method includes receiving, by the data processing system from a third user, a second request to conduct a second transaction with the second user…The method can also include generating, based on the determination, an indication of a high level of risk associated with the second transaction”, broadest reasonable interpretation of a request to determine a risk assessment includes a request where a risk level is determined for the request)
At the time of filing, one of ordinary skill in the art would have motivation to combine Huang, Vinayagasundaram, Barddal, Gama, and Mislove by taking the ensemble of decision tree for risk prediction, where nodes related to depreciated variables are pruned, as jointly taught by Huang, Vinayagasundaram, Barddal, and Gama, and using the method to produce a risk prediction for a second request associated with a second user, as taught by Mislove, as it is well-known within the art that a robust method for prediction is able to accommodate multiple requests for a prediction to be made, and which imparts the predictable benefit of a wider range of uses for the method. Such a combination would be obvious.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Jordan et al. (U.S. Patent Application Publication No. 2021/0174264) teaches the training and use of tree-based machine learning models, including random forests, for applications such as risk prediction.
Steele et al. (U.S. Patent No. 10,339,465) teaches techniques for optimizations during the training of decision tree models.
Globerson and Roweis “Nightmare at Test Time: Robust Learning by Feature Deletion” teaches techniques to render machine learning classifiers robust against deletion of entire features.
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 VICTOR A NAULT whose telephone number is (703) 756-5745. The examiner can normally be reached M - F, 12 - 8.
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 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.
/V.A.N./Examiner, Art Unit 2124
/MIRANDA M HUANG/ Supervisory Patent Examiner, Art Unit 2124
Prosecution Insights