US 11,948,149 B1
Systems and methods for virtual certification number source verification
Kiran Kumar Vallabhaneni, Dublin, CA (US); Nishant Garg, Fremont, CA (US); Kelly Jo Brown, Rockville, MD (US); Jesse Emery, Arlington, VA (US); Jonathan Blocksom, Reston, VA (US); Victoria Yang, fAIRFAX, VA (US); Brian Glowniak, mClEAN, VA (US); Edward Husa, mClEAN, VA (US); and Calvin Sun, mClEAN, VA (US)
Assigned to Capital One Services, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Nov. 23, 2022, as Appl. No. 18/058,535.
Claims priority of provisional application 63/374,693, filed on Sep. 6, 2022.
Int. Cl. G06Q 20/40 (2012.01); G06N 3/088 (2023.01); G06N 3/09 (2023.01)
CPC G06Q 20/401 (2013.01) [G06N 3/088 (2013.01); G06N 3/09 (2023.01)] 14 Claims
OG exemplary drawing
 
1. A method for authorizing a source for virtual certification number (VCN) use, the method comprising:
receiving a VCN use request from an external entity, the VCN use request comprising a VCN and transaction data associated with the VCN use request;
determining that a VCN binding for the VCN is a strict binding;
training an authorization convolutional neural network model by:
(i) parsing historical transaction data into a plurality of components based on fields associated with the historical transaction data,
(ii) extracting a first hash value from the plurality of components,
(iii) outputting each of a first transaction source identifier that includes the extracted first hash value, and a first authorization confidence score for the first transaction source identifier, and
(iv) adjusting one or more authorization convolutional neural network model nodes, weights, layers, or biases, based at least in part on a comparison of the first authorization confidence score to a known authorization confidence score;
providing the transaction data to the trained authorization convolutional neural network model;
receiving, from the trained authorization convolutional neural network model, an output comprising a second authorization confidence score for a second transaction source identifier that includes an extracted second hash value, the output being associated with the transaction data;
mapping the transaction data to a transaction source having the second transaction source identifier based on the VCN binding being the strict binding and based on the second authorization confidence score for the second transaction source identifier meeting an authorization confidence score threshold;
determining a VCN source identifier for a merchant source associated with the VCN, based on a registration authentication of the VCN with the merchant source, wherein the registration authentication is based on registration confidence score output by a registration convolutional neural network model and wherein the registration convolutional neural network model is trained by adjusting one or more registration convolutional neural network model nodes, weights, layers, or biases based on historical determinations of registration or a likelihood of registration;
comparing the VCN source identifier to the second transaction source identifier;
determining that the VCN source identifier corresponds to the second transaction source identifier based on comparing the VCN source identifier to the second transaction source identifier; and
outputting an approved authorization based on determining that the VCN source identifier corresponds to the second transaction source identifier, the approved authorization indicating that the VCN is associated with the source and the source is the transaction source.