Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Priority
The present application has a provisional application No. 63/496,844 filed on April 18, 2023.
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 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 1 is a method claim thus it falls into one of the four categories of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding independent claim 1, following limitations recite a judicial exception:
“determining a simple rule for each of a plurality of parties based on transactional data possessed by each party”
[Mental Process] – determining a simple rule based on transactional data possessed by each party simply requires to compare the data to come up with the rule for each party which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“encoding the simple rule for each party using a local bloom filter that is specific for each party”
[Mental Process] – encoding a simple rule for each party using a filter is equivalent to simply creating a rule for each party that the filter can be used to find a match which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“merging all local bloom filters of the plurality of parties by an aggregator through federated learning to generate a global bloom filter”
[Mental Process] – merging all local bloom filters to generate a global bloom filter is simply adding all the filters created locally and combine them to make a filter that encompasses all the rules which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“removing intrinsic anomalies from the transactional data by the aggregator using the global bloom filter to obtain an augmented dataset of the transactional data”
[Mental Process] – removing anomalies using the global filter to obtain augmented dataset basically means using the simple rule filter to filter out data that does not match which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“determining a classifier by the aggregator based on the augmented dataset”
[Mental Process] – determining a classifier based on the dataset is simply require to compare the dataset to choose a classifier to be used which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“identifying complex anomalies using the classifier”
[Mental Process] – identifying anomalies using a classifier basically goes through the data to find out anomaly that is hard to notice which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 1, the claim recites additional elements of
“federated learning” and “classifier”
These AI related terms are recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 2
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 2 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 2 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 2, the claim recites additional elements of
“garbled circuit”
The circuit is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 3
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 3 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 3 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 3, the claim recites additional elements of
“XGBoost”
XGBoost is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 4
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 4 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 4, following limitations recite a judicial exception:
“the step of determining the classifier comprises augmenting the transactional data with account-level features”
[Mental Process] – augmenting the transactional data with account-level features is simply adding additional information to the data which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 4, the claim recites additional elements of
“classifier”
The classifier is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 5
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 5 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 5 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 5 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 6
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 6 is a dependent claim of 5, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 6 does not have any abstract idea by itself, thus uses all the limitations of Claim 5.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 6 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 7
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 7 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 7 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 7 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 8
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 8 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 8 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 8, the claim recites additional elements of
“rule-based classifier”
The classifier is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 9
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 9 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 9, following limitations recite a judicial exception:
“the step of determining the simple rule comprises encrypting the transactional data”
[Mental Process] – encrypting the transactional data requires to compute and convert the data into unreadable format, which can be any format meaning it can possibly be personalized format which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 9 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 10
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 10 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 10, following limitations recite a judicial exception:
“the intrinsic anomalies have an anomaly ratio greater than or equal to a threshold value”
[Mathematical Relations] – having ratio grater than or equal to a certain numeric value is simply stating it does have a mathematical relations which recites to an abstract idea
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 10 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 11
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 11 is a dependent claim of 10, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 11 does not have any abstract idea by itself, thus uses all the limitations of Claim 10.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 11 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 12
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 12 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 12 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 12 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 13
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 13 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 13 does not have any abstract idea by itself, thus uses all the limitations of Claims Y and Z.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 13, the claim recites additional elements of
“server devices”
The devices are recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 14
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 14 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 14 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 14, the claim recites additional elements of
“the classifier comprises a model based on linear regression, logistic regression, decision trees, support vector machines (SVM), naive Bayes, k-nearest neighbors or K-nearest neighbors (k-NN), K-means clustering, random forest, dimensionality reduction algorithms, gradient boosting algorithms, or neural networks”
The classifier and its types are recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 15
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 15 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 15 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 15, the claim recites additional elements of
“the classifier comprises one or more machine learning models”
The machine learning models are recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 16
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 16 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
As Claim 16 does not have any abstract idea by itself, thus uses all the limitations of Claim 1.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 16, the claim recites additional elements of
“the classifier comprises a neural network, a convolutional neural network (CNN), a deep convolutional neural network (DCNN), a cascaded deep convolutional neural network, a simplified CNN, a shallow CNN, or a combination thereof”
The classifier and its’ type of neural networks are recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 17
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 17 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 17, following limitations recite a judicial exception:
“determining a simple rule for each of a plurality of parties based on transactional data possessed by each party”
[Mental Process] – Reason1 which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
[Mental Process] – determining a simple rule based on transactional data possessed by each party simply requires to compare the data to come up with the rule for each party which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“encoding the simple rule for each party using a local bloom filter that is specific for each party”
[Mental Process] – encoding a simple rule for each party using a filter is equivalent to simply creating a rule for each party that the filter can be used to find a match which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“merging all local bloom filters of the plurality of parties by an aggregator through federated learning to generate a global bloom filter”
[Mental Process] – merging all local bloom filters to generate a global bloom filter is simply adding all the filters created locally and combine them to make a filter that encompasses all the rules which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“removing intrinsic anomalies from the transactional data by the aggregator using the global bloom filter to obtain an augmented dataset of the transactional data”
[Mental Process] – removing anomalies using the global filter to obtain augmented dataset basically means using the simple rule filter to filter out data that does not match which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
“determining a classifier by the aggregator based on the augmented dataset”
[Mental Process] – determining a classifier based on the dataset is simply require to compare the dataset to choose a classifier to be used which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 17, the claim recites additional elements of
“federated learning” and “classifier”
These AI related terms are recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
“training a classifier by the aggregator based on the augmented dataset”
Training a classifier using dataset is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional elements do no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional elements [1,2] are considered a mere instruction to apply an exception to the generic computer components or machine-learning components that simply run mathematical calculations and mental processes (see MPEP 2106.05(f)). This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 18, 19, 20
Claims 18-20 have similar limitations of Claims 2-4, respectively. For the reasons described above with respect to Claims 2-4, these judicial exceptions are not meaningfully integrated into a practical application, or significantly more than the abstract ideas. The claims do not provide anything more than the abstract ideas of mental processes and mathematical calculations that are practically capable of being performed with the assistance of pen and paper. Therefore, Claims 18-20 also recite abstract ideas that do not integrate into a practical application or amount to significantly more than judicial exception, and thus are rejected under U.S.C. 101.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 2, 8, 9, 13-18 are rejected under 35 U.S.C. 103 as being unpatentable over Niu et al. (Niu), Non-Patent Literature, “Secure Federated Submodel Learning”, published in Nov 2019, arXiv, 29 Pages, in view of Al-Dalky et al. (Al), Non-Patent Literature, “Accelerating Snort NIDS using NetFPGA-based Bloom Filter”, published on 2014, IEEE, 6 Pages, in further view of Banerjee et al. (Banerjee), Non-Patent Literature, “Identity Management with Hybrid Blockchain Approach: A Deliberate Extension with Federated-Inverse-Reinforcement Learning”, Published on 2021, IEEE, 6 Pages.
As to independent Claim 1,
Niu teaches a method for anomaly identification through privacy-enhanced two-step federated learning, comprising:
determining for each of a plurality of parties based on data possessed by each party (
Niu, Pg11, Right Column, Paragraph1, Lines10-12, "A chosen client determines her real index set based on her local data, which can specify the “position” of her truly required submodel",
Pg11, Algorithm1, Lines10, "Determines her real index set S(i) based on local data;",
Pg5, Table1, Definition of S(i), Lines9-10, "Client i’s real index set that corresponds to local data and specifies truly required rows of W", wherein distributed clients act as a plurality of parties that independently determine their required submodel index sets S(i) based on their locally stored user feature data, which corresponds to determining a selection criteria for each of a plurality of parties based on data possessed by each party);
encoding for each party using a local bloom filter that is specific for each party (
Niu, Pg10, Left Column, Paragraph2, Lines10-11, "We first let each chosen client represent her real index set as a Bloom filter"
Pg8, Right Column, Section 3), Paragraph2, Lines5-7, "To represent a set of elements, we apply h hash functions to each element and set the Bloom filter at the positions of hash values to 1"
Pg14, Algorithm3 , Lines3, "Represents S(i) as a Bloom filter b(i)", wherein Niu discloses encoding a client's local private dataset S(i) into a local Bloom filter b(i) by setting bit position to 1 using h hash functions, rendering it functionally equivalent to the claimed invention of encoding a criterion for each party using a local bloom filter);
merging all local bloom filters of the plurality of parties by an aggregator through federated learning to generate a global bloom filter (
Niu, Pg10, Right Column, Lines2-5, "we let the cloud server directly “sum” the Bloom filters. Here, the sum operation can be performed obliviously and efficiently under the coordination of an untrusted cloud server with the secure aggregation protocol"
Pg9, Left Column, Paragraph2, Lines2-7, "To represent sets..., we use Bloom filters,.. b(i). Then the union of these sets i.e.,
PNG
media_image1.png
25
85
media_image1.png
Greyscale
, can be represented by a Bloom filter, which performs bitwise OR operations over the Bloom filters, i.e.,
PNG
media_image2.png
29
84
media_image2.png
Greyscale
"
Pg14, Algorithm3, Lines8,
PNG
media_image3.png
46
487
media_image3.png
Greyscale
, wherein Niu discloses that each client represents its local set as a Bloom filter b(i), and the cloud server (aggregator) aggregates all local Bloom filters via bitwise OR operations or secure aggregation protocols to generate a global union Bloom filter representing the combined dataset, rendering it functionally equivalent to the claimed invention);
removing intrinsic anomalies from the data by the aggregator using the global bloom filter to obtain an augmented dataset of the data (
Niu, Pg8, Section 3), Paragraph2, Lines7-11, "In the membership test phase, to check whether an element belongs to the set, we simply check the Bloom filter at the positions of its hash values. If h any of the bits at these positions is 0, the element is definitely not in the set",
Pg15, Left Column Lines9-11, "By simply doing membership tests for the indices falling into these partitions, the cloud server can efficiently construct the union",
Pg21, Left Column, Paragraph2, Lines6-12, "extracts her succinct training set from the original training set by following two rules: (1) For the goods ID to be predicted in a sample, if it does not belong to the succinct set of goods IDs, this sample will be filtered out; and (2) for the sequence of historical goods IDs in a sample, we only keep those goods IDs in the succinct set of goods IDs as well as their corresponding category IDs"
Algorithm1, Line14,
PNG
media_image4.png
61
492
media_image4.png
Greyscale
Pg12, Left Column, Lines15-18, "each client can augment the matrix, denoting her weighted submodel update, with the transposed count vector in the last column, when preparing materials for secure aggregation (Line 19)"
Pg11, Right Column, Paragraph1, Lines44-46, "adding the update of the succinct submodel to the rows with the succinct indices and padding zero vectors to the other rows (Line 16)", wherein Niu explicitly teaches using a global Bloom filter to conduct membership testing and filtering out non-matching data entries to prepare a refined training dataset. Regarding the removal of 'intrinsic anomalies' via global Bloom filter querying, Niu explicitly teaches querying a global Bloom filter to evaluate set membership and filtering out non-compliant data entries from a dataset prior to model training. Under the Broadest Reasonable Interpretation, 'intrinsic anomalies' encompasses predetermined rule-violating or non-member data entries detectable via lookup mechanisms. Niu also teaches obtaining an augmented dataset/matrix through feature appending and vector padding based on Bloom filter lookup results to construct the expanded dataset, rendering it functionally equivalent to the claimed invention of filtering/removing the dataset using the global bloom filter to construct an augmented dataset.)
Niu, however, does not explicitly teach
determining a simple rule based on data
simple rules to be used for encoding
From the same field of endeavor, Al teaches determining a simple rule based on data (Al, Pg872, Left Column, Section B, Lines6-15, "The function monitor_rules() which will be implemented as a part of Snort NIDS software and periodically monitor the rules set in Snort NIDS along with the offloaded rules on NetFPGA... After finding the most frequent rules to be offloaded, monitor_rules() will call the parser() function that will extract the 5-tuples from the most frequent rules that will be written to NetFPGA",
Pg871, Left Column, Last Paragraph, Lines3-5, "The Header parser extracts 5-tuples from the packet’s header. The 5-tuples are: IP source and destination, addresses, source and destination ports and protocol id", Pg871, Right Column, Paragraph2, Lines1-6, “Let’s have the following Snort rule: alert tcp any any ->any any, the hash value corresponding to this rule is 6 as the any tuples’ values are considered 0. This means that the bloom filter will have a value of 1 at index 6. Assume that we have an incoming packet with the following 5-tuples: tcp, 10.0.0.1, 10.0.0.2, 80, and 8080”, wherein Al teaches dynamically identifying and selecting specific filtering rules via monitor_rules() corresponds to determining a simple rule based on the data which the most frequent rules to be offloaded such that if Al uses the data from Niu’s each party from a plurality of parties. Also, Al teaches using hash values as simple rules to filter out matching packets. As Niu mentioned about encoding for each party using the bloom filter, if these determined simple rules by Al are encoded by the Niu’s bloom filter, it is functionally equivalent to the claimed invention.)
Al further reinforces the idea of removing intrinsic anomalies (Al, Pg3, Left Column, Section A, Lines7-16, "NetFPGA is considered as a first line of defense that will match the incoming network packets against the most frequently triggered rules... However, if the packet matches one of the offloaded rules, the NetFPGA will drop the packet (it will not be passed to Snort)...", wherein Niu teaches performing membership tests to filter out non-compliant data, Al explicitly discloses dropping or removing intrinsic anomaly that match or violate the simple-rule determined above, rendering the combination functionally equivalent to the claimed invention.)
Niu and Al are analogous to the claimed invention as they are from the same field of endeavor of privacy-preserving federated learning and anomaly detection in distributed data networks. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the local Bloom filter encoding and secure federated aggregation of Niu with Bloom filter-based pre-filtering of rule-based traffic of Al. The motivation is as recited by Al (Al, Pg869, Abstract, Lines22-25, “The experimental results show a significant improvement in the CPU usage and an enormous reduction in packet loss when using Snort with NetFPGA filtering”) such that offloading preliminary rule matching to Bloom filters prior to model processing, thereby reducing CPU computational overhead and improving overall processing throughput on the aggregator side.
However, both Niu and Al do not teach:
transactional data
determining a classifier by the aggregator based on the augmented dataset; and
identifying complex anomalies using the classifier.
From the same field of endeavor, Banerjee teaches these limitations. The combination of Niu and Al teaches that determining a simple rule for each of a plurality of parties based on data possessed by each party but is silent that the data is transactional data. Banerjee teaches transactional data (Banerjee, Pg1, Abstract, Lines17-19, "Our proposed method achieves the target of identity classifications from variety of transaction data")
Banerjee further teaches determining a classifier by the aggregator based on the augmented dataset (Banerjee, Pg2, Left Column, First Paragraph, Lines3-5, "a federated machine learning to train the decision engine with all the pieces of device identities"
Pg3, Right Column, Paragraph4, Lines4-13, "FL(federated learning) trains a shared global model by iteratively aggregating model updates from multiple client devices... Initially, eligible client devices first check-in with a remote server... train the model on their local datasets, and report their respective model updates to the server for aggregation"
Pg5, Right Column, Paragraph3, Lines7-13, "then we can easily classify the transactions based on their node and transaction features. In this context, the simulation is presented in Figure 2. As depicted in this figure, it is quite evident that the IRL(inverse reinforcement learning) master model of normal maximum likelihood outperforms, while selecting precisely the level of trust and honesty of devices as opposed to the IRL with SVM(support vector machine) as a master model", wherein Banerjee explicitly teaches training and determining a central decision engine/classifier (such as an IRL master model utilizing supervised classification algorithms like SVM or Maximum Likelihood Estimation) at a remote server or aggregator based on aggregated transaction and node dataset features, rendering it functionally equivalent to the claimed invention.)
Banerjee further teaches identifying complex anomalies using the classifier (Banerjee, Pg1, Abstract, Lines15-17, "In this context, it is important to identify and classify the malicious and non-malicious types of transactions"
Pg2, Left Column, Lines20-21, "...solicits more adaptive and intelligent algorithm to classify malicious and non-malicious types of transactions"
Pg4, 5, Left Column, Paragraph1, Lines2-6, "The dataset contains a large set of transactions, each of which is either labelled as licit (exchanges, wallet providers, miners, licit services, etc.), illicit (scams, malware, terrorist organizations, ransomware, Ponzi schemes, etc.) or unknown data instances"
Pg5, Right Column, Paragraph1, Lines7-17, "The paper considers these features: Total number of features : 166, Transaction Features (local features) : 94 (first), Nodes Features (Aggregated features) : 72 (remaining)... we have implemented Inverse Reinforcement Learning using different supervised classification algorithms, to help the IRL model learn an appropriate policy, which can accurately prevent occurrence of malicious/invalid transactions"
Pg5, Right Column, Paragraph3, Lines7-9, "then we can easily classify the transactions based on their node and transaction features", wherein Banerjee explicitly teaches using the trained decision engine/classifier to classify multi-dimensional, complex transaction data (comprising 166 combined local and aggregated feature) into licit versus illicit types (such as scams, malware, ransomware, and Ponzi schemes), which corresponds under the Broadest Reasonable Interpretation to identifying complex anomalies using the classifier, rendering it functionally equivalent to the claimed invention.)
Niu, Al and Banerjee are analogous to the claimed invention as they are from the same field of endeavor of secure federated learning, identify validation, and malicious transaction detection in distributed network systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the local Bloom filter encoding and secure federated aggregation of Niu and the preliminary Bloom filter rule-filtering of Al with the user/device identify classification and malicious transaction identification of Banerjee. The motivation is as recited by Banerjee (Banerjee, Pg2, Left Column, Lines19-21, "In fact, the decentralized mechanism through Blockchain solicits more adaptive and intelligent algorithm to classify malicious and non-malicious types of transactions") such that combining these features to augment transaction data with account-level identity metadata, enabling an intelligent classifier to accurately identify complex anomalies and validate account statuses across distributed parties without relying on a centralized third party.
As to dependent Claim 2,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
Niu further teaches the method of claim 1, wherein the step of merging is performed by garbled circuits (Niu, Pg14, Left Column, Lines32-44, "For the protocols based on Bloom filter the union operation over sets is normally transformed to the element-wise OR operation over Bloom filters, as demonstrated in Section III-B3, whereas the logic OR operation can be further converted to bit addition and bit multiplication. To obliviously perform addition and multiplication operations, the above two kinds of protocols mainly turn to generic secure two-party/multiparty computation (e.g., garbled circuit, homomorphic encryption, secret sharing, and oblivious transfer), or outsource secure computation to multiple non-colluding servers. Due to unaffordable computation and communication overheads, none of the existing private set union protocols have been deployed in practice", wherein Niu explicitly mentions that garbled circuit is one way of merging the filters, rendering it functionally equivalent to the claimed invention.)
As to dependent Claim 8,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
Niu, however, does not teach the following limitation, but from the same field of endeavor Al teaches the method of claim 1, wherein
the simple rule comprises a rule-based classifier (
Al, Pg869, Introduction, Lines10-13, "Signature-based detection matches the traffic against well-known attacks known as signatures which are represented as pre-defined pre-configured rules"
Pg870, Left Column, Section A, Paragraph3, Lines6-10, "The Detection Engine is the most important component of Snort. It is responsible to detect malicious packets. This can be achieved by checking packets against thousands of rules that represent well-known attack signatures"
Pg872, Left Column, Section B, Lines3-15, "Snort API has a number of functions that are responsible to manage the offloading process, update the Bloom filter and the counters corresponding to the offloaded rules. The function monitor_rules() which will be implemented as a part of Snort NIDS software and periodically monitor the rules set in Snort NIDS along with the offloaded rules on NetFPGA. This function will find the most frequent rules to be offloaded and the rules that need to be replaced within NetFPGA. After finding the most frequent rules to be offloaded, monitor_rules() will call the parser() function that will extract the 5-tuples from the most frequent rules that will be written to NetFPGA", wherein Al explicitly discloses utlizing Snort NIDS, which operates as a classic rule-based classifier. Snort's detection engine uses pre-defined, pre-configured signature rules to evaluate and classify network traffic into normal or malicious packets. The offloaded rules extracted from Snort consist of simple, deterministic header-matching criteria (5-tuples) designed to catch known signature patterns before forwarding unmatched traffic for deeper inspection. The disclosure employs a rule-based classification engine to establish simple rules that are subsequently encoded into a Bloom filter for fast preliminary filtering, rendering it functionally equivalent to the claimed invention)
Niu and Al are analogous to the claimed invention as they are from the same field of endeavor of privacy-preserving federated learning and anomaly detection in distributed data networks. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the local Bloom filter encoding and secure federated aggregation of Niu with Bloom filter-based pre-filtering of rule-based traffic of Al. The motivation is as recited by Al (Al, Pg869, Abstract, Lines22-25, “The experimental results show a significant improvement in the CPU usage and an enormous reduction in packet loss when using Snort with NetFPGA filtering”) such that offloading preliminary rule matching to Bloom filters prior to model processing, thereby reducing CPU computational overhead and improving overall processing throughput on the aggregator side.
As to dependent Claim 9,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
Niu further teaches the method of claim 1, wherein
the step of determining the simple rule comprises encrypting the transactional data (
Niu, Pg8, Left Column, Paragraph1, Lines7-12, "In addition, to guarantee confidentiality and integrity against the mediate cloud server, client-to-client messages should be encrypted with symmetric authenticated encryption, where the secret key is set up through Diffie-Hellman key exchange between two clients"
Pg20, Lines14-16, "the authenticated encryption used Advanced Encryption Standard (AES) in the Cipher Block Chaining (CBC) mode..."
Pg14, Algorithm3, Line4, "Perturbs b(i) to an integer vector b'(i) by replacing each bit 1 in b(i) with a random integer from Z_R", wherein Niu discloses protecting transaction-derived index sets using symmetric authenticated encryption (AES in CBC mode with Diffie-Hellman key exchange) and cryptographic perturbation over the Bloom filter vectors (b'(i) mapped to random integers in Z_R), which each local client (the corresponding party or bank) applies cryptographic encryption/masking prior to federated aggregation, rendering it functionally equivalent to the claimed invention.)
As to dependent Claim 13,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
Niu further teaches the method of claim 1, wherein
the aggregator is implemented on one or more server devices (
Niu, Pg2, Figure1 (b),
PNG
media_image5.png
109
352
media_image5.png
Greyscale
Pg6, Fig2, Line3, "where the cloud server, as the aggregator, only obtains the sum of vectors from multiple clients...", wherein Niu explicitly discloses the cloud server as the aggregator, rendering it functionally equivalent to the claimed invention.)
As to dependent Claim 14,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
Niu further teaches the method of claim 1, wherein
the classifier comprises a model based on linear regression, logistic regression, decision trees, support vector machines (SVM), naive Bayes, k-nearest neighbors or K-nearest neighbors (k-NN), K-means clustering, random forest, dimensionality reduction algorithms, gradient boosting algorithms, or neural networks (
Niu, Pg1, Right Column, Lines3-15, "In addition, the deployed recommendation models follow a golden paradigm of embedding and Multi-Layer Perceptron (MLP)... Deep Interest Network (DIN) [4] introduces the attention mechanism... Deep Interest Evolution Network (DIEN) further extracts latent interests and monitors interest evolution through Gated Recurrent Unit (GRU)... and Behavior Sequence Transformer (BST) [6] incorporates transformer"
Pg5, Left Column, Paragraph1, Lines5-7, "In particular, the query suggestion used logistic regression as the triggering model for on-device training", wherein Niu explicitly discloses that the classifier types are deep neural network architectures as described, rendering it functionally equivalent to the claimed invention.)
As to dependent Claim 15,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
Niu further teaches the method of claim 1, wherein
the classifier comprises one or more machine learning models (
Niu, Pg1, Right Column, Lines3-15, "In addition, the deployed recommendation models follow a golden paradigm of embedding and Multi-Layer Perceptron (MLP)... Deep Interest Network (DIN) [4] introduces the attention mechanism... Deep Interest Evolution Network (DIEN) further extracts latent interests and monitors interest evolution through Gated Recurrent Unit (GRU)... and Behavior Sequence Transformer (BST) [6] incorporates transformer"
Pg5, Left Column, Paragraph1, Lines5-7, "In particular, the query suggestion used logistic regression as the triggering model for on-device training", wherein Niu explicitly discloses the classifier types include one or more of machine learning models, rendering it functionally equivalent to the claimed invention.)
As to dependent Claim 16,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
Niu further teaches the method of claim 1, wherein
the classifier comprises a neural network, a convolutional neural network (CNN), a deep convolutional neural network (DCNN), a cascaded deep convolutional neural network, a simplified CNN, a shallow CNN, or a combination thereof (
Niu, Pg1, Right Column, Lines3-15, "In addition, the deployed recommendation models follow a golden paradigm of embedding and Multi-Layer Perceptron (MLP)... Deep Interest Network (DIN) [4] introduces the attention mechanism... Deep Interest Evolution Network (DIEN) further extracts latent interests and monitors interest evolution through Gated Recurrent Unit (GRU)... and Behavior Sequence Transformer (BST) [6] incorporates transformer", wherein Niu explicitly discloses the classifier types that includes a neural network such as MLP, rendering it functionally equivalent to the claimed invention.)
As to dependent Claim 17,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
Niu further teaches the method of claim 1, wherein the classifier is trained by:
determining for each of a plurality of parties based on data possessed by each party (
Niu, Pg11, Right Column, Paragraph1, Lines10-12, "A chosen client determines her real index set based on her local data, which can specify the “position” of her truly required submodel",
Pg11, Algorithm1, Lines10, "Determines her real index set S(i) based on local data;",
Pg5, Table1, Definition of S(i), Lines9-10, "Client i’s real index set that corresponds to local data and specifies truly required rows of W", wherein distributed clients act as a plurality of parties that independently determine their required submodel index sets S(i) based on their locally stored user feature data, which corresponds to determining a selection criteria for each of a plurality of parties based on data possessed by each party);
encoding for each party using a local bloom filter that is specific for each party (
Niu, Pg10, Left Column, Paragraph2, Lines10-11, "We first let each chosen client represent her real index set as a Bloom filter"
Pg8, Right Column, Section 3), Paragraph2, Lines5-7, "To represent a set of elements, we apply h hash functions to each element and set the Bloom filter at the positions of hash values to 1"
Pg14, Algorithm3 , Lines3, "Represents S(i) as a Bloom filter b(i)", wherein Niu discloses encoding a client's local private dataset S(i) into a local Bloom filter b(i) by setting bit position to 1 using h hash functions, rendering it functionally equivalent to the claimed invention of encoding a criterion for each party using a local bloom filter);
merging all local bloom filters of the plurality of parties by an aggregator through federated learning to generate a global bloom filter (
Niu, Pg10, Right Column, Lines2-5, "we let the cloud server directly “sum” the Bloom filters. Here, the sum operation can be performed obliviously and efficiently under the coordination of an untrusted cloud server with the secure aggregation protocol"
Pg9, Left Column, Paragraph2, Lines2-7, "To represent sets..., we use Bloom filters,.. b(i). Then the union of these sets i.e.,
PNG
media_image1.png
25
85
media_image1.png
Greyscale
, can be represented by a Bloom filter, which performs bitwise OR operations over the Bloom filters, i.e.,
PNG
media_image2.png
29
84
media_image2.png
Greyscale
"
Pg14, Algorithm3, Lines8,
PNG
media_image3.png
46
487
media_image3.png
Greyscale
, wherein Niu discloses that each client represents its local set as a Bloom filter b(i), and the cloud server (aggregator) aggregates all local Bloom filters via bitwise OR operations or secure aggregation protocols to generate a global union Bloom filter representing the combined dataset, rendering it functionally equivalent to the claimed invention);
removing intrinsic anomalies from the data by the aggregator using the global bloom filter to obtain an augmented dataset of the data (
Niu, Pg8, Section 3), Paragraph2, Lines7-11, "In the membership test phase, to check whether an element belongs to the set, we simply check the Bloom filter at the positions of its hash values. If h any of the bits at these positions is 0, the element is definitely not in the set",
Pg15, Left Column Lines9-11, "By simply doing membership tests for the indices falling into these partitions, the cloud server can efficiently construct the union",
Pg21, Left Column, Paragraph2, Lines6-12, "extracts her succinct training set from the original training set by following two rules: (1) For the goods ID to be predicted in a sample, if it does not belong to the succinct set of goods IDs, this sample will be filtered out; and (2) for the sequence of historical goods IDs in a sample, we only keep those goods IDs in the succinct set of goods IDs as well as their corresponding category IDs"
Algorithm1, Line14,
PNG
media_image4.png
61
492
media_image4.png
Greyscale
Pg12, Left Column, Lines15-18, "each client can augment the matrix, denoting her weighted submodel update, with the transposed count vector in the last column, when preparing materials for secure aggregation (Line 19)"
Pg11, Right Column, Paragraph1, Lines44-46, "adding the update of the succinct submodel to the rows with the succinct indices and padding zero vectors to the other rows (Line 16)", wherein Niu explicitly teaches using a global Bloom filter to conduct membership testing and filtering out non-matching data entries to prepare a refined training dataset. Regarding the removal of 'intrinsic anomalies' via global Bloom filter querying, Niu explicitly teaches querying a global Bloom filter to evaluate set membership and filtering out non-compliant data entries from a dataset prior to model training. Under the Broadest Reasonable Interpretation, 'intrinsic anomalies' encompasses predetermined rule-violating or non-member data entries detectable via lookup mechanisms. Niu also teaches obtaining an augmented dataset/matrix through feature appending and vector padding based on Bloom filter lookup results to construct the expanded dataset, rendering it functionally equivalent to the claimed invention of filtering/removing the dataset using the global bloom filter to construct an augmented dataset.)
Niu, however, does not explicitly teach
determining a simple rule based on data
simple rules to be used for encoding
From the same field of endeavor, Al teaches determining a simple rule based on data (Al, Pg872, Left Column, Section B, Lines6-15, "The function monitor_rules() which will be implemented as a part of Snort NIDS software and periodically monitor the rules set in Snort NIDS along with the offloaded rules on NetFPGA... After finding the most frequent rules to be offloaded, monitor_rules() will call the parser() function that will extract the 5-tuples from the most frequent rules that will be written to NetFPGA",
Pg871, Left Column, Last Paragraph, Lines3-5, "The Header parser extracts 5-tuples from the packet’s header. The 5-tuples are: IP source and destination, addresses, source and destination ports and protocol id", Pg871, Right Column, Paragraph2, Lines1-6, “Let’s have the following Snort rule: alert tcp any any ->any any, the hash value corresponding to this rule is 6 as the any tuples’ values are considered 0. This means that the bloom filter will have a value of 1 at index 6. Assume that we have an incoming packet with the following 5-tuples: tcp, 10.0.0.1, 10.0.0.2, 80, and 8080”, wherein Al teaches dynamically identifying and selecting specific filtering rules via monitor_rules() corresponds to determining a simple rule based on the data which the most frequent rules to be offloaded such that if Al uses the data from Niu’s each party from a plurality of parties. Also, Al teaches using hash values as simple rules to filter out matching packets. As Niu mentioned about encoding for each party using the bloom filter, if these determined simple rules by Al are encoded by the Niu’s bloom filter, it is functionally equivalent to the claimed invention.)
Al further reinforces the idea of removing intrinsic anomalies (Al, Pg3, Left Column, Section A, Lines7-16, "NetFPGA is considered as a first line of defense that will match the incoming network packets against the most frequently triggered rules... However, if the packet matches one of the offloaded rules, the NetFPGA will drop the packet (it will not be passed to Snort)...", wherein Niu teaches performing membership tests to filter out non-compliant data, Al explicitly discloses dropping or removing intrinsic anomaly that match or violate the simple-rule determined above, rendering the combination functionally equivalent to the claimed invention.)
Niu and Al are analogous to the claimed invention as they are from the same field of endeavor of privacy-preserving federated learning and anomaly detection in distributed data networks. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the local Bloom filter encoding and secure federated aggregation of Niu with Bloom filter-based pre-filtering of rule-based traffic of Al. The motivation is as recited by Al (Al, Pg869, Abstract, Lines22-25, “The experimental results show a significant improvement in the CPU usage and an enormous reduction in packet loss when using Snort with NetFPGA filtering”) such that offloading preliminary rule matching to Bloom filters prior to model processing, thereby reducing CPU computational overhead and improving overall processing throughput on the aggregator side.
However, both Niu and Al do not teach:
transactional data
training a classifier by the aggregator based on the augmented dataset
The combination of Niu and Al teaches that determining a simple rule for each of a plurality of parties based on data possessed by each party but is silent that the data is transactional data. From the same field of endeavor, Banerjee teaches the transactional data (Banerjee, Pg1, Abstract, Lines17-19, "Our proposed method achieves the target of identity classifications from variety of transaction data")
Banerjee further teaches training a classifier by the aggregator based on the augmented dataset (Banerjee, Pg2, Left Column, First Paragraph, Lines3-5, "a federated machine learning to train the decision engine with all the pieces of device identities"
Pg3, Right Column, Paragraph4, Lines4-13, "FL(federated learning) trains a shared global model by iteratively aggregating model updates from multiple client devices... Initially, eligible client devices first check-in with a remote server... train the model on their local datasets, and report their respective model updates to the server for aggregation"
Pg5, Right Column, Paragraph3, Lines7-13, "then we can easily classify the transactions based on their node and transaction features. In this context, the simulation is presented in Figure 2. As depicted in this figure, it is quite evident that the IRL(inverse reinforcement learning) master model of normal maximum likelihood outperforms, while selecting precisely the level of trust and honesty of devices as opposed to the IRL with SVM(support vector machine) as a master model", wherein Banerjee explicitly teaches training a central decision engine/classifier (such as an IRL master model utilizing supervised classification algorithms like SVM or Maximum Likelihood Estimation) at a remote server or aggregator based on aggregated transaction and node dataset features, rendering it functionally equivalent to the claimed invention.)
Niu, Al and Banerjee are analogous to the claimed invention as they are from the same field of endeavor of secure federated learning, identify validation, and malicious transaction detection in distributed network systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the local Bloom filter encoding and secure federated aggregation of Niu and the preliminary Bloom filter rule-filtering of Al with the user/device identify classification and malicious transaction identification of Banerjee. The motivation is as recited by Banerjee (Banerjee, Pg2, Left Column, Lines19-21, "In fact, the decentralized mechanism through Blockchain solicits more adaptive and intelligent algorithm to classify malicious and non-malicious types of transactions") such that combining these features to augment transaction data with account-level identity metadata, enabling an intelligent classifier to accurately identify complex anomalies and validate account statuses across distributed parties without relying on a centralized third party.
As to dependent Claim 18,
it is a claim that contains similar limitations of Claim 2 and thus rejected under the same rationale.
Claims 3-7, 10-12, 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Niu, Al and Banerjee as mentioned in Claim 1, in further view of Yadav et al. (Yadav), US Patent Application No. US-2022/0164798-A1, Published in May 2022.
As to dependent Claim 3,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
However, the combination does not teach the following limitation, but from the same field of endeavor, Yadav teaches the method of claim 1, wherein
the step of determining the classifier is performed by training the classifier with XGBoost (
Yadav, Pg2, Paragraph27, Lines1-2, "In some embodiments, an XGBoost machine learning model may be used"
Pg6, Paragraph122, Lines1-5, "At step 660 , the enriched transaction data may be scored using a trained XGBoost Model. An XGBoost Model, is a decision-tree-based ensemble Machine Learning algorithm that uses a gradient boosting framework. An XGBoost software library may be used to train a model", wherein Yadav explicitly discloses the deployment of an XGBoost machine learning model, specifically detailing that an XGBoost software library is utilized to train the classification model for scoring and detecting anomalous transactions.)
Niu, Al, Banerjee and Yadav are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based anomaly and fraud detection in distributed transaction and network data processing systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the privacy-preserving federated submodel learning framework utilizing Bloom filter encoding and randomized response of Liu, the Bloom filter-based fast data filtering and feature matching technique of Al, and the consensus-guided federated learning framework across distributed client nodes of Banerjee with the graph-based and statistical feature extraction and enrichment framework for machine learning classifier scoring of Yadav. The motivation is as recited by Yadav (Yadav, Pg2, Paragraph26, Lines8-12, "Single-hub and multi-hub features extracted from graph provides important historic information about the transaction sender, recipient and their interaction for the machine learning model to identify fraudulent transaction patterns") such that incorporating graph-based structural and time-series interaction features into a privacy-enhanced federated classifier enables the system to capture both local proximity and complex multi-hop connectivity patterns among transaction participants, thereby significantly improving the detection accuracy and robustness against sophisticated anomalous and fraudulent behaviors in distributed financial payment networks without exposing sensitive client data.
As to dependent Claim 4,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
However, the combination does not teach the following limitation, but from the same field of endeavor, Yadav teaches the method of claim 1, wherein
the step of determining the classifier comprises augmenting the transactional data with account-level features (
Yadav, Pg3, Paragraph43, Lines3-7, "To augment this information to be able to train a good classifier , these transactions are enriched 204 with relevant features for the sender and recipient extracted from their past (historical) transaction behaviour", wherein Yadav explciitly discloses training classifier involves using transaction data that has been enriched or augmented with the account-level information, rendering it functionally equivalent to the claimed invention.)
Niu, Al, Banerjee and Yadav are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based anomaly and fraud detection in distributed transaction and network data processing systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the privacy-preserving federated submodel learning framework utilizing Bloom filter encoding and randomized response of Liu, the Bloom filter-based fast data filtering and feature matching technique of Al, and the consensus-guided federated learning framework across distributed client nodes of Banerjee with the graph-based and statistical feature extraction and enrichment framework for machine learning classifier scoring of Yadav. The motivation is as recited by Yadav (Yadav, Pg2, Paragraph26, Lines8-12, "Single-hub and multi-hub features extracted from graph provides important historic information about the transaction sender, recipient and their interaction for the machine learning model to identify fraudulent transaction patterns") such that incorporating graph-based structural and time-series interaction features into a privacy-enhanced federated classifier enables the system to capture both local proximity and complex multi-hop connectivity patterns among transaction participants, thereby significantly improving the detection accuracy and robustness against sophisticated anomalous and fraudulent behaviors in distributed financial payment networks without exposing sensitive client data.
As to dependent Claim 5,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
However, the combination does not teach the following limitation, but from the same field of endeavor, Yadav teaches the method of claim 1, wherein
the parties comprise banks and the aggregator comprises a financial institute (
Yadav, Pg5, Paragraph117, "In some embodiments, data may be enriched using multiple sources of transaction details. For example, transaction details from two or more financial institutions or other external payment channels may be collected as part of the historical transaction data. In some embodiments, multiple sources of real - time transaction details may be obtained and analysed. For example, a first financial institution may receive data feeds from a second financial institution (and/or from an external payment channel) to collect and store historical transaction details that may be used to increase the scope of constructed transaction graphs"
Pg5, Paragraph118, "In an operational example , if a first financial institution client X sends email money transfer to client Y from another financial institution , and client Y then sends money to another first financial institution client Z , then the first financial institution internal data alone will not be able to capture the indirect connection between the first financial institution clients X and Z. By using an other data source (e.g., an external payment channel details), the first financial institution would be able form such indirect connections between the client X and Z while constructing the transaction graph", wherein Yadav explicitly discloses obtaining transaction data from two or more financial institutions, first and second financial institutions, participating in a shared payment/transaction network. These participating institutions exchanging data corresponds to the claimed parties comprising banks. Yadav teaches that a first financial institution receives data feeds from other financial institutions to collect, aggregate, and store combined transaction details for graph construction analysis. The central collecting entity performing the aggregation is identified as a first financial institution corresponds to the aggregator being a financial institution, rendering it functionally equivalent to the claimed invention.)
Niu, Al, Banerjee and Yadav are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based anomaly and fraud detection in distributed transaction and network data processing systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the privacy-preserving federated submodel learning framework utilizing Bloom filter encoding and randomized response of Liu, the Bloom filter-based fast data filtering and feature matching technique of Al, and the consensus-guided federated learning framework across distributed client nodes of Banerjee with the graph-based and statistical feature extraction and enrichment framework for machine learning classifier scoring of Yadav. The motivation is as recited by Yadav (Yadav, Pg2, Paragraph26, Lines8-12, "Single-hub and multi-hub features extracted from graph provides important historic information about the transaction sender, recipient and their interaction for the machine learning model to identify fraudulent transaction patterns") such that incorporating graph-based structural and time-series interaction features into a privacy-enhanced federated classifier enables the system to capture both local proximity and complex multi-hop connectivity patterns among transaction participants, thereby significantly improving the detection accuracy and robustness against sophisticated anomalous and fraudulent behaviors in distributed financial payment networks without exposing sensitive client data.
As to dependent Claim 6,
The combination of Niu, Al, Banerjee and Yadav teaches, as mentioned above, all the limitations of Claim 5. It teaches that the federated learning uses banks/parties as edge source and a financial institute as an aggregator to collect data from the edge sources to train the classifier.
However, the combination of Niu, Al, and Banerjee does not teach the following limitation, but from the same field of endeavor, Yadav teaches the method of claim 5, wherein
the transactional data comprises bank transactional data (
Yadav, Pg1, Paragraph3, Lines1-2, "One of the services offered in online banking is an email money transfer between individuals"
Pg3, Paragraph42, Lines6-7, "...historical transactional data 310 (e.g., historical email transfer data)"
Pg5, Table1,
PNG
media_image6.png
271
444
media_image6.png
Greyscale
, wherein Yadav explicitly discloses bank transactional data as shown in Table 1, rendering it functionally equivalent to the claimed invention.)
Niu, Al, Banerjee and Yadav are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based anomaly and fraud detection in distributed transaction and network data processing systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the privacy-preserving federated submodel learning framework utilizing Bloom filter encoding and randomized response of Liu, the Bloom filter-based fast data filtering and feature matching technique of Al, and the consensus-guided federated learning framework across distributed client nodes of Banerjee with the graph-based and statistical feature extraction and enrichment framework for machine learning classifier scoring of Yadav. The motivation is as recited by Yadav (Yadav, Pg2, Paragraph26, Lines8-12, "Single-hub and multi-hub features extracted from graph provides important historic information about the transaction sender, recipient and their interaction for the machine learning model to identify fraudulent transaction patterns") such that incorporating graph-based structural and time-series interaction features into a privacy-enhanced federated classifier enables the system to capture both local proximity and complex multi-hop connectivity patterns among transaction participants, thereby significantly improving the detection accuracy and robustness against sophisticated anomalous and fraudulent behaviors in distributed financial payment networks without exposing sensitive client data.
As to dependent Claim 7,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
However, the combination does not teach the following limitation, but from the same field of endeavor, Yadav teaches the method of claim 1, wherein
the transactional data comprises account information (
Yadav, Pg3, Paragraph43, Lines1-3, "Each transaction comprises information about the sender and recipient of the transaction along with the amount of money being transferred"
Pg5, Table1,
PNG
media_image6.png
271
444
media_image6.png
Greyscale
, wherein Yadav explicitly discloses that each transaction contains detailed information regarding the sender and recipient of the transaction, as well as the transferred amount. Furthermore, Table 1 enumerates exact data fields of these transactions, such as client card numbers, recipient card numbers, and payee account identifiers, rendering it functionally equivalent to the claimed invention.)
Niu, Al, Banerjee and Yadav are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based anomaly and fraud detection in distributed transaction and network data processing systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the privacy-preserving federated submodel learning framework utilizing Bloom filter encoding and randomized response of Liu, the Bloom filter-based fast data filtering and feature matching technique of Al, and the consensus-guided federated learning framework across distributed client nodes of Banerjee with the graph-based and statistical feature extraction and enrichment framework for machine learning classifier scoring of Yadav. The motivation is as recited by Yadav (Yadav, Pg2, Paragraph26, Lines8-12, "Single-hub and multi-hub features extracted from graph provides important historic information about the transaction sender, recipient and their interaction for the machine learning model to identify fraudulent transaction patterns") such that incorporating graph-based structural and time-series interaction features into a privacy-enhanced federated classifier enables the system to capture both local proximity and complex multi-hop connectivity patterns among transaction participants, thereby significantly improving the detection accuracy and robustness against sophisticated anomalous and fraudulent behaviors in distributed financial payment networks without exposing sensitive client data.
As to dependent Claim 10,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
However, the combination does not teach the following limitation, but from the same field of endeavor, Yadav teaches the method of claim 1, wherein
the intrinsic anomalies have an anomaly ratio greater than or equal to a threshold value (
Yadav, Pg6, Paragraph125, Lines1-11, "In some embodiments, rules may be toggled or otherwise suspended if too many false positives are found to be present . A false positive ratio per rulemay be measured. An example of a false positive ratio is (the number of false positive alerts per rule) / (the number of true positive alerts). In some embodiments, when the false positive ratio per rule is over a set threshold, then the system may suspend that rule and allow the transaction to proceed... If the ratio lowers below the set threshold, then that rule may be reinforced", wherein Yadav discloses the concept of evaluating transaction anomaly rules by comparing a calculated rule-based ratio against a predefined threshold value. The claimed invention defines intrinsic anomalies as transaction conditions where an anomaly ratio meets or exceeds an established threshold value to classify transactions under rule-based filtering, while Yadav similarly calculates a ratio of anomalous or false positive occurrences per rule and determines whether that ratio exceeds or falls below a set threshold to dynamically control rule application, rendering it functionally equivalent to the claimed invention of computing a quantitative ratio of anomalous behaviors relative to transaction alerts and executing rule-based classification actions when the ratio satisfies a specified threshold condition.)
Niu, Al, Banerjee and Yadav are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based anomaly and fraud detection in distributed transaction and network data processing systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the privacy-preserving federated submodel learning framework utilizing Bloom filter encoding and randomized response of Liu, the Bloom filter-based fast data filtering and feature matching technique of Al, and the consensus-guided federated learning framework across distributed client nodes of Banerjee with the graph-based and statistical feature extraction and enrichment framework for machine learning classifier scoring of Yadav. The motivation is as recited by Yadav (Yadav, Pg2, Paragraph26, Lines8-12, "Single-hub and multi-hub features extracted from graph provides important historic information about the transaction sender, recipient and their interaction for the machine learning model to identify fraudulent transaction patterns") such that incorporating graph-based structural and time-series interaction features into a privacy-enhanced federated classifier enables the system to capture both local proximity and complex multi-hop connectivity patterns among transaction participants, thereby significantly improving the detection accuracy and robustness against sophisticated anomalous and fraudulent behaviors in distributed financial payment networks without exposing sensitive client data.
As to dependent Claim 11,
The combination of Niu, Al, Banerjee and Yadav teaches, as mentioned above, all the limitations of Claim 10. It teaches that the intrinsic anomalies have anomaly ratio greater than or equal to a threshold.
However, the combination of Niu, Al, and Banerjee does not teach the following limitation, but from the same field of endeavor, Yadav teaches the method of claim 10, wherein
the threshold value is 0.95 (
Yadav, Pg6, Paragraph125, Lines1-11, "In some embodiments, rules may be toggled or otherwise suspended if too many false positives are found to be present . A false positive ratio per rule may be measured. An example of a false positive ratio is (the number of false positive alerts per rule) / (the number of true positive alerts). In some embodiments, when the false positive ratio per rule is over a set threshold, then the system may suspend that rule and allow the transaction to proceed... If the ratio lowers below the set threshold, then that rule may be reinforced", wherein Yadav discloses evaluating anomaly rules by comparing a calculated ratio against a predefined threshold value. Although Yadav does not explicitly teach the specific numerical value of 0.95 for the threshold, selecting and optimizing a specific numerical threshold such as 0.95, as a matter of routine experimentation and design choice depending on the targeted sensitivity, precision, and false-positive tolerance of the system. Modifying or selecting specific numerical limits within a known range or framework to achieve optimal filtering efficiency is a result-effective variable. Furthermore, the specification of the claimed invention does not demonstrate any critical or unexpected technical effect resulting specifically from setting the threshold at 0.95. Therefore, it renders obvious over Yadav in view of routine engineering optimization.)
Niu, Al, Banerjee and Yadav are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based anomaly and fraud detection in distributed transaction and network data processing systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the privacy-preserving federated submodel learning framework utilizing Bloom filter encoding and randomized response of Liu, the Bloom filter-based fast data filtering and feature matching technique of Al, and the consensus-guided federated learning framework across distributed client nodes of Banerjee with the graph-based and statistical feature extraction and enrichment framework for machine learning classifier scoring of Yadav. The motivation is as recited by Yadav (Yadav, Pg2, Paragraph26, Lines8-12, "Single-hub and multi-hub features extracted from graph provides important historic information about the transaction sender, recipient and their interaction for the machine learning model to identify fraudulent transaction patterns") such that incorporating graph-based structural and time-series interaction features into a privacy-enhanced federated classifier enables the system to capture both local proximity and complex multi-hop connectivity patterns among transaction participants, thereby significantly improving the detection accuracy and robustness against sophisticated anomalous and fraudulent behaviors in distributed financial payment networks without exposing sensitive client data.
As to dependent Claim 12,
The combination of Niu, Al and Banerjee teaches, as mentioned above, all the limitations of Claim 1. It teaches the overall architecture of federated learning using local Bloom filters to encode simple-rules which the filters will be merged by the aggregator of the federated learning to rule-filtering to obtain augmented datasets to train a classifier by the aggregator to detect complex anomalies.
However, the combination does not teach the following limitation, but from the same field of endeavor, Yadav teaches the method of claim 1, wherein
the complex anomalies comprise statistical anomalies (
Yadav, Pg1, Paragraph5, Lines2-7, "The method of detecting fraudulent electronic transactions. The method transaction data, extracting graph-based and statistical features to enrich the real-time transaction data, and determining an account proximity score for the real-time transaction data"
Pg1, Paragraph23, Lines1-3, "It is desirable to detect anomalous email money transfers, based on known fraud cases using graph technology and supervised machine learning model"
Pg5, Paragraph105-108, "In some embodiments, time series features include: xcn_tmstmp_unix_lag_diff_max : the maximum of time interval (in seconds) between consecutive historical transactions from source to destination, xcn_tmstmp_unix_lag_diff_min : the minimum of time interval (in seconds) between consecutive historical transactions from source to destination, xcn_tmstmp_unix_lag_diff_avg : the average of time interval (in seconds) between historical transactions from source to destination"
Pg2, Paragraph27, Lines1-2, "In some embodiments, an XGBoost machine learning model may be used", wherein Yadav explicitly discloses detecting anomalous transaction patterns using a machine learning classifier trained on statistical features extracted from historical transaction data. Specifically, Yadav teaches enriching real-time transaction data with statistical metrics - such as transaction amount sums, interaction counts, and average time intervals - to compute a fraud probability score via an XGBoost model. Because these anomalous transactions are identified through the statistical distribution and temporal variance of transaction behaviors rather than simple hardcoded rules, Yadav's detected fraud cases directly corresponds to complex anomalies that comprises statistical anomalies, rendering it functionally equivalent to the claimed invention.)
Niu, Al, Banerjee and Yadav are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based anomaly and fraud detection in distributed transaction and network data processing systems. Therefore, it would have been obvious to one of ordinary skills in the art, before the effective filing date, to combine the privacy-preserving federated submodel learning framework utilizing Bloom filter encoding and randomized response of Liu, the Bloom filter-based fast data filtering and feature matching technique of Al, and the consensus-guided federated learning framework across distributed client nodes of Banerjee with the graph-based and statistical feature extraction and enrichment framework for machine learning classifier scoring of Yadav. The motivation is as recited by Yadav (Yadav, Pg2, Paragraph26, Lines8-12, "Single-hub and multi-hub features extracted from graph provides important historic information about the transaction sender, recipient and their interaction for the machine learning model to identify fraudulent transaction patterns") such that incorporating graph-based structural and time-series interaction features into a privacy-enhanced federated classifier enables the system to capture both local proximity and complex multi-hop connectivity patterns among transaction participants, thereby significantly improving the detection accuracy and robustness against sophisticated anomalous and fraudulent behaviors in distributed financial payment networks without exposing sensitive client data.
As to dependent Claim 19,
it is a claim that contains similar limitations of Claim 3 and thus rejected under the same rationale.
As to dependent Claim 20,
it is a claim that contains similar limitations of Claim 4 and thus rejected under the same rationale.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DONG YOON JUNG whose telephone number is (571)270-0198. The examiner can normally be reached 8am-5pm.
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, Cesar Paula can be reached at (571) 272-4128. 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.
/DONG YOON JUNG/ Examiner, Art Unit 2145
/CESAR B PAULA/Supervisory Patent Examiner, Art Unit 2145
Prosecution Insights