Systems, methods, and computer program products for determining uncertainty from a deep learning classification model
Disclosed are systems for determining uncertainty from a deep learning classification model. A system for determining uncertainty from a deep learning classification model may include at least one processor programmed or configured to determine a classification for an input based on a deep learning classification model, generate an uncertainty score for the classification, determine whether the uncertainty score satisfies a threshold, in response to determining that the uncertainty score satisfies the threshold, determine an automated action based on a decision model, and in response to determining that the uncertainty score does not satisfy the threshold, determine the automated action based on at least one predefined rule. Methods and computer program products are also disclosed.
1 . A system comprising:
a transaction processing system arranged in an electronic payment processing network with at least one issuer system and a plurality of merchants, the transaction processing system comprising at least one processor programmed or configured to:
receive, through the electronic payment processing network, a stand-in processing request for a requested transaction from a merchant based on the issuer system being offline or unavailable to authorize the requested transaction;
determine a classification for an input based on a deep learning classification model, the input comprising transaction data for the requested transaction;
generate an uncertainty score for the classification based on an output of the deep learning classification model;
determine whether the uncertainty score satisfies a threshold;
in response to determining that the uncertainty score satisfies the threshold, determine an automated action based on a decision model, the automated action comprising authorizing or rejecting the stand-in processing request; and
in response to determining that the uncertainty score does not satisfy the threshold, determine the automated action based on at least one predefined rule from rules data independent of the deep learning classification model, the automated action comprising authorizing or rejecting the stand-in processing request.
2 . The system of claim 1 , wherein the uncertainty score comprises a confidence interval based on a predicted probability.
3 . The system of claim 1 , wherein the at least one processor is programmed or configured to generate the uncertainty score for the classification as an output of the deep learning classification model.
4 . The system of claim 1 , wherein the at least one processor is programmed or configured to:
determine two parameters of a Beta distribution of predicted probabilities based on the deep learning probability model, wherein generating the uncertainty score for the classification is based on the two parameters.
5 . The system of claim 1 , wherein the at least one processor is programmed or configured to:
determine two parameters of a logit layer of the deep learning probability model, wherein generating the uncertainty score for the classification is based on the two parameters.
6 . The system of claim 1 , wherein the input is received from a client computer in communication with the transaction processing system.
7 . A method comprising:
receiving, by at least one processor of a transaction processing system arranged in an electronic payment processing network with at least one issuer system and a plurality of merchants, a stand-in processing request for a requested transaction from a merchant based on the issuer system being offline or unavailable to authorize the requested transaction;
determining, with at least one processor, a classification for an input based on a deep learning classification model, the input comprising transaction data for the requested transaction;
generating, with at least one processor, an uncertainty score for the classification based on an output of the deep learning classification model;
determining, with at least one processor, whether the uncertainty score satisfies a threshold; and
in response to determining that the uncertainty score satisfies the threshold, determining, with at least one processor, an automated action based on a decision model from rules data independent of the deep learning classification model, the automated action comprising authorizing or rejecting the stand-in processing request.
8 . The method of claim 7 , further comprising:
in response to determining that the uncertainty score does not satisfy the threshold, determining the automated action based on at least one predefined rule.
9 . The method of claim 7 , further comprising:
generating the uncertainty score for the classification as an output of the deep learning classification model.
10 . The method of claim 7 , further comprising:
determining two parameters of a Beta distribution of predicted probabilities based on the deep learning probability model, wherein generating the uncertainty score for the classification is based on the two parameters.
11 . The method of claim 7 , further comprising:
determining two parameters of a logit layer of the deep learning probability model, wherein generating the uncertainty score for the classification is based on the two parameters.
12 . A computer program product comprising at least one non-transitory computer-readable medium including program instructions that, when executed by at least one processor of a transaction processing system arranged in an electronic payment processing network with at least one issuer system and a plurality of merchants, cause the at least one processor to:
receive, through the electronic payment processing network, a stand-in processing request for a requested transaction from a merchant based on the issuer system being offline or unavailable to authorize the requested transaction;
determine a classification for an input based on a deep learning classification model, the input comprising transaction data for the requested transaction;
generate an uncertainty score for the classification based on an output of the deep learning classification model;
determine whether the uncertainty score satisfies a threshold;
in response to determining that the uncertainty score satisfies the threshold, determine an automated action based on a decision model, the automated action comprising authorizing or rejecting the stand-in processing request; and
in response to determining that the uncertainty score does not satisfy the threshold, determine the automated action based on at least one predefined rule from rules data independent of the deep learning classification model, the automated action comprising authorizing or rejecting the stand-in processing request.
13 . The computer program product of claim 12 , wherein the uncertainty score comprises a confidence interval based on a predicted probability.
14 . The computer program product of claim 12 , wherein the program instructions further cause the at least one processor to:
generate the uncertainty score for the classification as an output of the deep learning classification model.
15 . The computer program product of claim 12 , wherein the program instructions further cause the at least one processor to:
determine two parameters of a Beta distribution of predicted probabilities based on the deep learning probability model, wherein generating the uncertainty score for the classification is based on the two parameters.
16 . The computer program product of claim 12 , wherein the program instructions further cause the at least one processor to:
determine two parameters of a logit layer of the deep learning probability model, wherein generating the uncertainty score for the classification is based on the two parameters.
17 . The computer program product of claim 12 , wherein the input comprises transaction data for a requested transaction.
18 . The computer program product of claim 12 , wherein the input is received from a client computer in communication with the transaction processing system.