DETAILED ACTION
This Final Office Action is in response Applicant communication filed on
2/3/2026. In Applicant’s amendment, claims 1, 10, and 19 were amended.
Claims 1-20 are currently pending and have been rejected as follows.
Response to Amendments
Rejections under 35 USC 101 are maintained. Applicant’s amendments necessitated new grounds of rejection under 35 USC 103.
Response to Arguments
Applicant’s 35 USC 101 rebuttal arguments and amendments have been fully considered but they are not persuasive to overcome the rejection.
Applicant argues on p. 11 that the claims are not directed to certain methods of organizing human activity because the claimed features models data as a heterogenous knowledge graph and data modeling with relationships and interactions between entities and because determining whether the first node and second node are fraudulent to predict a transaction cannot amount to mere organizing of human activity. Examiner respectfully disagrees. Data modeling to compute a score indicating a likelihood of an occurrence of a transaction falls squarely under certain methods of organizing human activity like commercial interactions (including sales activities or behaviors; business relations) and mathematical concepts like mathematical calculations. Limiting information analysis to particular real-world content does not change its informational character and make a claim nonabstract.
Applicant argues on p. 11 that the claims recite a practical application because the claims offer improved accuracy in predicting whether a future transaction will occur based on the transaction features and relationship features and based on the prediction made on if the first node or second node is fraudulent. Examiner respectfully disagrees. Under Step 2A, Prong 2, examiners should evaluate whether the claim as a whole integrates the recited judicial exception into a practical application of the exception. Limitations the courts have found indicative that an additional element (or combination of elements) may have integrated the exception into a practical application include:
An improvement in the functioning of a computer, or an improvement to other technology or technical field, as discussed in MPEP §§ 2106.04(d)(1) and 2106.05(a);
Applying or using a judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition, as discussed in MPEP § 2106.04(d)(2);
Implementing a judicial exception with, or using a judicial exception in conjunction with, a particular machine or manufacture that is integral to the claim, as discussed in MPEP § 2106.05(b);
Effecting a transformation or reduction of a particular article to a different state or thing, as discussed in MPEP § 2106.05(c); and
Applying or using the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception, as discussed in MPEP § 2106.05(e).
The courts have also identified limitations that did not integrate a judicial exception into a practical application:
Merely reciting the words "apply it" (or an equivalent) with the judicial exception, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f);
Adding insignificant extra-solution activity to the judicial exception, as discussed in MPEP § 2106.05(g); and
Generally linking the use of a judicial exception to a particular technological environment or field of use, as discussed in MPEP § 2106.05(h).
Here, the claimed additional elements and judicial exception are not an improvement to a computer or a technology. The claim language requires embeddings to be generated based in part on the set of node features and the set of edge features and on the first set of node embeddings and the second set of node embeddings. There is no recitation of a technical mechanism by which the computer or neural network is improved. The alleged integration is an improvement to the accuracy of a sales transaction prediction metric. Improved metric accuracy is an improvement to the abstract idea. The claims do not recite an improvement to the function of a computer, or an improvement to other technology or technical field. The claims merely use computers as tools to perform the abstract idea.
Applicant argues on p. 12 that the claims are eligible due to their alleged similarity to claims 2 and 3 of Example 35 because claims 2 and 3 recited fraud detection and were held to have practical applications carried out in a non-conventional and non-generic way. Examiner respectfully disagrees. Claims 1-3 of Example 35 were all characterized as directed to the abstract idea of fraud prevention through identity verification. Claims 2 and 3 of Example 35 recited a specific sequence that differs from the routine and conventional sequence of events normally conducted by ATM verification to address the unique problems associated with bank cards and ATMs. Claim 2 included the use of generating an image containing encrypted code data, transmitting it to a registered mobile communication device associated with the bank card, read and decrypted the code data from the image on the customer’s registered mobile communication device. Claim 3’s similar sequence included the ATM obtaining the random code transmitted from the customer’s mobile communication device and controlling access to the ATM in response. In contrast, the present claims produces a prediction on if the first node or second node is fraudulent and a score indicating a likelihood of an occurrence of a transaction between the first node and the second node. The present claims do not recite a specific coordinated interaction among particular machines, encryption/decryption, security code comparison, or access control responsive to the security sequence. The present claims are more analogous to ineligible claim 1 of Example 35, which obtained information, compared it, authenticated a customer, and determined whether the transaction should proceed when a match from the comparison verifies the authenticity of the customer’s identity.
Response to Arguments
Applicant’s prior art arguments and amendments have been fully considered but they are moot in light of the newly cited portions of the Wadha reference below.
In particular, see Wadha [0090] “The plurality of graph features may include, but not limited to, geolocation data associated with the financial transactions, population density, transaction velocity (i.e., frequency of financial transaction by a user to a particular user), historical fraud data, and transaction history. The historical fraud data may provide information of users who were engaged in fraud financial activities;” [0092] “At 425, the server system 106 generates a temporal knowledge graph based on the plurality of graph features. The temporal knowledge graph represents the one or more related users engaged in the financial transactions as related nodes and relations among the related nodes as edges. The edges may be, but not limited to, geolocation data associated with the financial transaction, social connection, and fraud connection” noting the two related users represented as nodes and the characteristics including historical fraud data and [0094] noting the processing of each edge; and [0100]-[0101] describing the time-based probabilities to a predict a connection between a particular node and a source node and updating a node fraud score of the particular node.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are clearly drawn to at least one of the four categories of patent eligible subject matter recited in 35 U.S.C. 101 (method, system, and non-transitory computer readable medium). Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without integrating the abstract idea into a practical application or amounting to significantly more than the abstract idea.
Regarding Step 1 of the 2019 Revised Patent Subject Matter Eligibility Guidance (‘2019 PEG”), Claims 1-9 are directed toward the statutory category of a process (reciting a “method”). Claims 10-18 are directed toward the statutory category of an article of manufacturer (reciting a “non-transitory computer readable medium”). Claims 19-20 are directed toward the statutory category of a machine (reciting a “system”).
Regarding Step 2A, prong 1 of the 2019 PEG, Claims 1, 10 and 19 are directed to an abstract idea by reciting accessing a graph comprising a plurality of nodes and a plurality of edges linking the plurality of nodes, the plurality of nodes comprising a first type of nodes representing a first type of entities and a second type of nodes representing a second type of entities; extracting a set of node features for each of the plurality of nodes; extracting a set of edge features for each of the plurality of edges; for an edge that connects a first node of the first type and a second node of the second type, generating (1) a first set of node embeddings for the first node and (2) a second set of node embeddings for the second node based in part on the set of node features and the set of edge features; generating a set of edge embeddings for the edge based in part on the first set of node embeddings, the second set of node embeddings, and the set of edge features, wherein the set of edge embeddings are generated from the edge of a heterogenous knowledge graph of a neural network, wherein the edge represents interactions and relationships between the first node and the second node, wherein characteristics of a connection between the first node and the second node are processed and a prediction is made on if the first node or second node is fraudulent, and wherein a prediction is made on whether a transaction will occur based on the interactions and the relationships associated with the edge; and computing a score based in part on the set of edge embeddings, the score indicating a likelihood of an occurrence of a transaction between the first node and the second node (Example Claim 1).
The claims are considered abstract because these steps recite certain methods of organizing human activity like commercial interactions (including sales activities or behaviors; business relations) and mathematical concepts like mathematical calculations. The claims focus on scoring the likelihood of an occurrence of a transaction between nodes by accessing a graph, extracting features, generating embeddings, and computing a score. Applicant’s disclosure describes the problem the claimed steps aim to solve as current machine learning models operating on tabular data failing to capture the complexity of the real world relationships between companies, customers, and salespersons (Applicant’s Specification, [0017]-[0019]). By this evidence, the claims recite a type of “certain methods of organizing human activity like commercial interactions (including sales activities or behaviors; business relations) and mathematical concepts like mathematical calculations” common to judicial exception to patent-eligibility. By preponderance, the claims recite an abstract idea (e.g., a “system” for classifying nodes or edges of graphs based on node embeddings and edge embeddings).
Regarding Step 2A, prong 2 of the 2019 PEG, the judicial exception is not integrated into a practical application because the claims (the judicial exception and the additional elements such as a processor; and a non-transitory computer readable storage medium) are not an improvement to a computer or a technology, the claims do not apply the judicial exception with a particular machine, the claims do not effect a transformation or reduction of a particular article to a different state or thing nor do the claims apply the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment such that the claims as a whole is more than a drafting effort designed to monopolize the exception (see MPEP §§ 2106.05(a-c, e)).
Claims 9 and 18 recite the use of a neural network at a high level. This merely amounts to using a computer as a tool to perform an abstract idea.
Dependent claims 2-9, 11-18, and 20 do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the limitations recite mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea ‐ see MPEP 2106.05(f).
Regarding Step 2B of the 2019 PEG, the additional elements have been considered above in Step 2A Prong 2. The claim limitations do not amount to significantly more than the judicial exception because they are directed to limitations referenced in MPEP 2106.05I.A. that are not enough to qualify as significantly more when recited in a claim with an abstract idea because the limitations recite mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea ‐ see MPEP
2106.05(f).
Applicant's claims mimic conventional, routine, and generic computing by their similarity to other concepts already deemed routine, generic, and conventional [Berkheimer Memorandum, Page 4, item 2] by the following [MPEP § 2106.05(d) Part (II)]. The claims recite steps like: “Receiving or transmitting data over a network, e.g., using the Internet to gather data,” Symantec, “Performing repetitive calculations,” Flook, and “storing and retrieving information in memory,” Versata Dev. Group, Inc. v. SAP Am., Inc. (citations omitted), by performing steps for “accessing” a graph, “extracting” a set of node features and a set of edge features, “generating” sets of embeddings, and “computing” a score (Example Claim 1).
By the above, the claimed computing “call[s] for performance of the claimed information collection, analysis, and display functions ‘on a set of generic computer components' and display devices” [Elec. Power Group, 830 F.3d at 1355] operating in a “normal, expected manner” [DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d at 1245, 1258 (Fed. Cir. 2014)].
Conclusively, Applicant's invention is patent-ineligible. When viewed both individually and as a whole, Claims 1-20 are directed toward an abstract idea without integration into a practical application and lacking an inventive concept.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 USC 103 as being unpatentable over the teachings of
Liu et al., US 20240054356 A1, hereinafter Liu, in view of
Zhang et al., US 20220343068 A1, hereinafter Zhang, in view of
Sarshogh et al., US 20210334896 A1, hereinafter Sarshogh, in view of
Wadha et al., US 20220020026 A1, hereinafter Wadha. As per,
Claims 1, 10, 19
Liu teaches
A computer-implemented method, the method comprising: /
A non-transitory computer-readable medium, stored thereon computer-executable instructions, that when executed by a processor of a computer system, cause the computer system to: /
A computer system comprising: a processor; and a non-transitory computer readable storage medium, stored thereon computer-executable instructions, that when executed by the processor, cause the processor to: (Liu [0035]-[0037])
accessing a graph comprising a plurality of nodes and a plurality of edges linking the plurality of nodes, the plurality of nodes comprising a first type of nodes representing a first type of entities and a second type of nodes representing a second type of entities; (Liu fig. 2A; [0017] “an example graph (e.g., representing an organization) comprising a plurality of nodes and edges is shown. A node may represent or otherwise indicate an employee, person, user, customer, team, product, software code repository, system, dataset, document, resource, project, or a variety of other entities or items.” Note the nodes and edges and multiple type of entities)
extracting a set of node features for each of the plurality of nodes; (Liu [0016] “The graph data may include one or more nodes that are representative of entities, documents, resources, users (e.g., customers), banking products, or a variety of other objects or concepts.” Corresponding to node features)
extracting a set of edge features for each of the plurality of edges; (Liu [0016] “An edge connecting two nodes in the graph may indicate a relationship, an event, or any other association between the two nodes.” Corresponding to edge features)
for an edge that connects a first node of the first type and a second node of the second type, generating (1) a first set of node embeddings for the first node and (2) a second set of node embeddings for the second node based in part on the set of node features and the set of edge features; (Liu [0004] “a system may generate, based on a graph including a plurality of nodes, a first set of node embeddings via a first embedding model and a second set of node embeddings via a second embedding model” noting the sets of generated node embeddings)
[…];
[…].
Liu does not explicitly teach, Zhang however in the analogous art of node and edge classifications teaches
generating a set of edge embeddings for the edge based in part on the first set of node embeddings, the second set of node embeddings, and the set of edge features, […]; (Zhang [0054] “If node I and j are directly connected, the edge embedding r.sub.ij is initialized as the type of dependency relation and fine-tuned during the training process. If node I and j are not directly connected, the edge embedding r.sub.ij is the sum of all the edge segments” corresponding to the generated set of edge embeddings)
computing a score based in part on the set of edge embeddings, the score indicating a likelihood of an occurrence of a transaction between the first node and the second node. (Zhang [0026] “Intent prediction, where the output of the graph encoding is used for predicting the intent action and object;” [0061] “Regarding the intent detection prediction, the output to fully connected layers is used separately for predicting the sequential labels of the intent in the following three categories: None, ACT and OBJ. Here, “ACT” represents the intent action and “OBJ” is the intent object. If a token doesn’t belong to neither intent action or object, they are categorized as the label “None.”” Noting the output results of None, ACT and OBJ corresponding to the computed score indicating a likelihood of an occurrence of a transaction)
Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to modify Liu’s node and edge classification to include generating edge embeddings and a predicted outcome in view of Zhang in an effort to improve a user’s sales, marketing, or customer service objectives (see Zhang ¶ [0075] & MPEP 2143G).
Liu / Zhang do not explicitly teach, Sarshogh however in the analogous art of node and edge classifications teaches
[…], wherein the set of edge embeddings are generated from the edge of a heterogenous knowledge graph of a neural network, wherein the edge represents interactions and relationships between the first node and the second node, and […]; (Sarshogh [0034] “The node aggregation model for customers may take, as input, the samples from its neighborhood (with importance sampling) and receive the current latent representation (e.g., embedding vectors 109) of each merchant. The node aggregation model may then aggregate these vectors with the vector 109 for the customer … the merchant graph neural network 105-1 and the customer graph neural network 105-2 may be included in the heterogenous graph neural network 105-3.” Note the embeddings representing relationships between the first node (merchant) and second node (customer) in a heterogenous graph neural network)
Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to modify Liu’s node and edge classification and Zhang’s edge embeddings to include a heterogenous graph neural network in view of Sarshogh in an effort to handle heterogeneity of the different data provided by the data sources (see Sarshogh ¶ [0034] & MPEP 2143G).
Liu / Zhang / Sarshogh do not explicitly teach, Wadha however in the analogous art of node and edge classifications teaches
[…], wherein characteristics of a connection between the first node and the second node are processed and a prediction is made on if the first node or second node is fraudulent, wherein a prediction is made on whether a transaction will occur based on the interactions and the relationships associated with the edge; and (Wadha [0090] “The plurality of graph features may include, but not limited to, geolocation data associated with the financial transactions, population density, transaction velocity (i.e., frequency of financial transaction by a user to a particular user), historical fraud data, and transaction history. The historical fraud data may provide information of users who were engaged in fraud financial activities;” [0092] “At 425, the server system 106 generates a temporal knowledge graph based on the plurality of graph features. The temporal knowledge graph represents the one or more related users engaged in the financial transactions as related nodes and relations among the related nodes as edges. The edges may be, but not limited to, geolocation data associated with the financial transaction, social connection, and fraud connection” noting the two related users represented as nodes and the characteristics including historical fraud data and [0094] noting the processing of each edge; and [0100]-[0101] describing the time-based probabilities to a predict a connection between a particular node and a source node and updating a node fraud score of the particular node; [0069] “with respect to financial transactions between two users, the training engine 218 may rely on a long short-term memory (LSTM) network (or other sequence neural network) to train the data model by consuming the real-time graph embedding vectors. Based on the trained data model, the LSTM network may predict next money laundering financial transactions;” [0071] “The prediction engine 220 is configured to determine time-based probabilities … The time-based probabilities may include, but not limited to, a time-based probability of next edge formation” note the prediction of a transaction occurring; [0077] “The plurality of graph features may include, but not limited to, geolocation data associated with the financial transactions, population density, transaction velocity (i.e., frequency of financial transaction among users),historical fraud data, and transaction history” note the transaction features)
Before the effective filing date of the claimed invention, it would have been obvious for one of ordinary skill in the art to modify Liu’s node and edge classification, Zhang’s edge embeddings, and Sarshogh’s heterogenous graph to include a transaction occurrence prediction and fraud prediction in view of Wadha in an effort to automate predictions for next financial transaction occurrence (see Wadha ¶ [0034] & MPEP 2143G).
Claims 2, 11, 20
Liu teaches
[…];
generating and sending a notification to an entity associated with the first node or the second node. (Liu [0035] “as shown in FIG. 3, both mobile device 322 and user terminal 324 include a display upon which to display data (e.g., conversational response, queries, and/or notifications).”)
Liu / Wadha / Sarshogh do not explicitly teach, Zhang however in the analogous art of node and edge classifications teaches
responsive to determining that the score is greater than a threshold, determining that the transaction between the first node and the second node is likely to occur; and (Zhang [0026] “Intent prediction, where the output of the graph encoding is used for predicting the intent action and object;” [0061] “Regarding the intent detection prediction, the output to fully connected layers is used separately for predicting the sequential labels of the intent in the following three categories: None, ACT and OBJ. Here, “ACT” represents the intent action and “OBJ” is the intent object. If a token doesn't belong to neither intent action or object, they are categorized as the label “None.”” Noting the output results of None, ACT and OBJ corresponding to the computed score indicating a likelihood of an occurrence of a transaction)
The motivations/rationales to combine Liu / Wadha / Sarshogh with Zhang persists.
Claims 3, 12
Liu teaches
traversing at least a subset of edges of the graph to compute a score for each edge in the subset; and (Liu [0042] “model 302 may include multiple layers (e.g., where a signal path traverses from front layers to back layers) … During testing, an output layer of model 302 may indicate whether or not a given input corresponds to a classification of model 302”)
visualizing at least a portion of the graph containing the subset of edges based in part on the computed scores. (Liu [0044] “System 300 also includes application programming interface (API) layer 350. API layer 350 may allow the system to generate summaries across different devices” corresponding to the visualizing of the graph data)
Claim 4
Liu teaches
identifying a subset of nodes that is within a neighborhood of a target node; and (Liu fig. 2A; [0020] “the computing system 102 may determine that two nodes are similar if a distance metric between their corresponding node embeddings is less than a threshold distance.”)
identifying the subset of edges that link the subset of nodes within the neighborhood. (Liu fig. 2A-2B; [0018] “edges connecting nodes 204-206 to each other may indicate that the users represented by nodes 204-206 conducted one or more transactions (e.g., a banking transaction, a blockchain transaction, etc.) with each other;” [0019] “node 220 may indicate a project and/or a product (e.g., a software product, banking product, etc.) associated with a transaction represented by node 202”)
Claims 5, 14
Liu teaches
wherein the first type of nodes represents salespersons, the second type of nodes represents leads, and the transaction between the first node and the second node is a conversion of a lead into a customer. (Liu [0017] “A node may represent or otherwise indicate an employee, person, user, customer, team” corresponding to a first type of note representing a salesperson and a second type representing lead/person/user/customer; [0018] “For example, edge 203 may be associated with a timestamp indicating a date and/or time that the user indicated by node 204 conducted a transaction with the user indicated by node 202” corresponding to the transaction between the two nodes)
Claims 6, 15
Liu teaches
wherein the subset of edges is within a neighborhood of the first node corresponding to a particular salesperson, and the subset of edges are visualized and presented to client device of the particular salesperson. (Liu fig. 2A; [0017] “A node may represent or otherwise indicate an employee” corresponding to a particular salesperson; [0044] “System 300 also includes application programming interface (API) layer 350. API layer 350 may allow the system to generate summaries across different devices” corresponding to the visualizing of the graph data)
Claims 7, 16
Liu teaches
wherein the first type of nodes represents merchants, the second type of nodes represents customers, and the transaction between the first node and the second node is a purchase transaction that a customer corresponding to the second node purchases a good or service from a merchant corresponding to the first node. (Liu [0017] “A node may represent or otherwise indicate an employee, person, user, customer, team, … or a variety of other entities or items” corresponding to a fist node type representing a merchant and a second node type representing a customer; [0018] “edges connecting nodes 204-206 to each other may indicate that the users represented by nodes 204-206 conducted one or more transactions (e.g., a banking transaction, a blockchain transaction, etc.) with each other” corresponding to a purchase of a good or service)
Claims 8, 17
Liu teaches
wherein the subset of edges is within a neighborhood of the first node corresponding to a particular merchant, and the subset of edges are visualized and presented to client device of the particular merchant. (Liu fig. 2A; [0017] “A node may represent or otherwise indicate an employee, person, user, customer, team, … or a variety of other entities” corresponding to a particular merchant; [0044] “System 300 also includes application programming interface (API) layer 350. API layer 350 may allow the system to generate summaries across different devices” corresponding to the visualizing of the graph data)
Claims 9, 18
Liu teaches
accessing a neural network model trained over data associated with the graph, wherein the neural network is trained to: (Liu [0022] “the computing system 102 may use GraphSAGE, GraphSAINT, a graph convolutional network” note the neural network model)
receive the set of node features and the set of edge features to generate the first set of node embeddings for the first node, the second set of node embeddings for the second node, and […]; (Liu [0017] “an example graph (e.g., representing an organization) comprising a plurality of nodes and edges is shown. A node may represent or otherwise indicate an employee, person, user, customer, team, product, software code repository, system, dataset, document, resource, project, or a variety of other entities or items. An edge (e.g., edge 203 or edge 230) may indicate an association between two nodes. An edge may indicate an event that two nodes were part of.” Note the node and edge features)
[…].
Liu / Wadha / Sarshogh do not explicitly teach, Zhang however in the analogous art of node and edge classifications teaches
[…] the edge embeddings for the edge; and (Zhang [0054] “If node i and j are directly connected, the edge embedding r.sub.ij is initialized as the type of dependency relation and fine-tuned during the training process. If node i and j are not directly connected, the edge embedding r.sub.ij is the sum of all the edge segments” corresponding to the generated set of edge embeddings)
computing the score based in part on the set of edge embeddings. (Zhang [0026] “Intent prediction, where the output of the graph encoding is used for predicting the intent action and object;” [0061] “Regarding the intent detection prediction, the output to fully connected layers is used separately for predicting the sequential labels of the intent in the following three categories: None, ACT and OBJ. Here, “ACT” represents the intent action and “OBJ” is the intent object. If a token doesn’t belong to neither intent action or object, they are categorized as the label “None.”” Noting the output results of None, ACT and OBJ corresponding to the computed score indicating a likelihood of an occurrence of a transaction)
The motivations/rationales to combine Liu / Wadha / Sarshogh with Zhang persists.
Claim 13
Liu teaches
wherein the subset of edges is within a neighborhood of a particular node. (Liu fig. 2A; [0020] “the computing system 102 may determine that two nodes are similar if a distance metric between their corresponding node embeddings is less than a threshold distance.”)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
US 2023/0334332 A1: Techniques are disclosed for performing adversarial object detection. In one example, a system obtains a feature vector upon receiving an object to be classified. The system then generates a graph using the feature vector for the object and other feature vectors that are respectively obtained from a reference set of objects, whereby the feature vector corresponds to a center node of the graph. The system uses a distance metric to select neighbor nodes from among the reference set of objects for inclusion into the graph, and then determines edge weights between nodes of the graph based on a distance between respective feature vectors between nodes. The system then applies a graph discriminator to the graph to classify the object as adversarial or benign, the graph discriminator being trained using (I) the feature vectors associated with nodes of the graph and (II) the edge weights between the nodes of the graph.
WO 2023/010502 A1: A method for training a Graph Neural Network (GNN) with multiple layers for anomaly detection is disclosed. The method comprising obtaining an embedding for each node of a graph and an embedding for the entire graph; inputting the embedding for each node and the embedding for the entire graph to the GNN, wherein the embedding for each node and the embedding for the entire graph are updated successively at each layer of the GNN; and updating the GNN based on a loss function.. The method further comprising edges of the graph are updated at each layer of the GNN before the embedding for each node and the embedding for the entire graph are updated. Numerous other aspects are provided.
Dipartimento et al., Fraud Prevention and Detection on Heterogeneous Information Networks with Deep Graph Infomax, 2021, Fraud is ubiquitous in both institutions and private companies, costing $1.9 billion in losses in the US, in 2019 alone. Fraud detection introduces a way to mitigate these losses while ensuring better security and enabling trust between all parties. However, it frequently comes at a great cost of resources needed to locate and oppose fraudulent cases manually. This cost also grows exponentially with the size of a company's financial transaction network. In this work, we propose a novel framework for automatic fraud detection that relies on institutions' readily available data, that aims at reducing the cost of resources outlined above. We evaluate our framework by comparing it against a baseline result and show an increase of 37% in performance expressed as F1-score while providing highly desirable characteristics such as online learning capability and a reduction of 82% in training time on commodity hardware.
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MOHAMED EL-BATHY whose telephone number is (571)270-5847. The examiner can normally be reached on M-F 8AM-4:30PM.
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, PATRICIA MUNSON can be reached on (571) 270-5396. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/MOHAMED N EL-BATHY/Primary Examiner, Art Unit 3624
Prosecution Insights