Self-Calibrating Fraud Detection
First Claim
Patent Images
1. A method for dynamically updating a model comprising:
- accessing a model that specifies expected characteristics for a transaction, wherein the model comprises variables associated with fraud;
receiving at least one value for each of the variables while monitoring transactions; and
updating a distribution of values for each variable based on the received value, wherein the received value is compared with the updated distribution to determine a deviation from a threshold value associated with a percentile of the updated distribution that is indicative of fraud.
1 Assignment
0 Petitions
Accused Products
Abstract
A method for dynamically updating a model is described. The method includes accessing a model that specifies expected characteristics for a transaction. The model includes variables associated with fraud. The method also includes receiving at least one value for each of the variables while monitoring transactions, and updating a distribution of values for each variable based on the received value. The received value is compared with the updated distribution to determine a deviation from a threshold value associated with a percentile of the updated distribution that is indicative of fraud.
-
Citations
30 Claims
-
1. A method for dynamically updating a model comprising:
-
accessing a model that specifies expected characteristics for a transaction, wherein the model comprises variables associated with fraud; receiving at least one value for each of the variables while monitoring transactions; and updating a distribution of values for each variable based on the received value, wherein the received value is compared with the updated distribution to determine a deviation from a threshold value associated with a percentile of the updated distribution that is indicative of fraud. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23)
-
-
24. A method for processing a transaction comprising:
-
receiving a value for a variable associated with a transaction, wherein the variable is indicative of fraud; updating a distribution of values for the variable based on the received value, wherein the updated distribution is used to determine a deviation of the received value from a selected value in the updated distribution; and determining a score indicative of a probability of fraud for the transaction based on the deviation of the received value. - View Dependent Claims (25)
-
-
26. A system for determining fraud comprising:
-
an interface to receive values associated with a transaction, each value corresponding to a property correlated to fraudulent transactions; a profile updater to modify a distribution of values for each property based on the corresponding received value; and a score calculator to generate a fraud score for the transaction, wherein generating the fraud score comprises comparing the received value with the updated distribution to determine a deviation from a threshold value of the updated distribution that is indicative of fraud.
-
-
27. A method for generating a fraud indicator comprising:
generating a score indicative of a probability of fraud for a transaction, the generating comprising aggregating self-scaling variables, wherein the self-scaling variables are determined by updating a distribution of values for each self-scaling variable with a received value for the self-scaling variable; determining an updated threshold value based on the updated distribution, wherein the updated threshold value indicates a beginning of a range of values that are unusual relative to the updated distribution; and scaling the received value based on the updated threshold value. - View Dependent Claims (28, 29)
-
30. A computer program product tangibly embodied in an information carrier, the computer program product including instructions that, when executed, perform operations for determining a fraud indicator, the operations comprising:
-
accessing a model that specifies expected characteristics for a transaction, wherein the model comprises variables associated with fraud; receiving at least one value for each of the variables while monitoring transactions; and updating a distribution of values for each variable based on the received value, wherein the received value is compared with the updated distribution to determine a deviation from a threshold value of the updated distribution that is indicative of fraud.
-
Specification