Image analysis – Applications – Personnel identification
Reexamination Certificate
1996-12-20
2002-08-06
Johnson, Timothy M. (Department: 2623)
Image analysis
Applications
Personnel identification
C382S124000, C382S156000, C382S228000, C705S044000, C340S005410, C340S005530, C340S005830, C902S005000
Reexamination Certificate
active
06430305
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the problem of providing electronic identification. More particularly, the present invention relates to methods for verifying the identity of a purchaser at a remote location for purposes of credit or cash debit transactions.
2. The Prior Art
There are numerous methods known in the prior art for remote verification of identity for various purposes, such as security, credit card transaction verification, automated teller machine transaction verification, etc. Such methods include identification of individual physical characteristics such as fingerprints and retinal patterns, and other indicia such as signature verification.
These prior art methods all attempt, with varying degrees of success, to limit access to cash, credit, to limit physical access to secure areas etc. At present, signature verification systems are not, by themselves, in general use for the purpose of making go
o-go decisions regarding the identity of a signing purchaser. This is in part caused by the fact that the accuracy rates of signature verification are currently too low to be commercially useful. The risk of offending legitimate customers by incorrectly identifying their transactions as fraudulent is too high when balanced against the number of successfully thwarted attempts at fraud resulting from correctly identifying transactions as fraudulent.
Some credit card issuers presently employ a variety of statistical models to detect patterns of suspicious credit card transactions. These statistical models vary widely, but generally examine such factors as the number of transactions, timing of those transactions, geographic locations of the transactions, type of merchant, purchase amounts, card type (e.g., “preferred” vs. “standard” card), historical information about the purchase patterns of individual customers, customer financial status, etc. The value of these statistical analyses is directly related to how quickly from the time at which a credit card is stolen a fraudulent transaction pattern can be identified. The fewer such transactions allowed to occur, the less money the credit card issuer ends up paying to cover them.
BRIEF DESCRIPTION OF THE INVENTION
According to the present invention, a signature verification technique is combined with additional indicia to arrive at a confidence value that a certain transaction is valid.
A method for fraud-pattern detection according to the present invention includes the steps of: providing a multi-layer neural network combiner having N “zero-one” inputs and 1 “zero-one” output; collecting training data for the combiner, including the steps of recording a plurality of signature verification attempts and their associated transactions, keeping track of the transactions which are fraudulent transactions; training the multi-layer neural network combiner on the N “zero-one” inputs and 1 “zero-one” output, wherein the N inputs are used to represent the probabilities output from each of the individual estimators which are to be combined using on-line back propagation with targets of zero for those transactions which turned out to be fraudulent and one for those transactions which turned out to be valid; and evaluating individual transactions, by first generating the individual probabilities, and then combining them by presenting the individual probabilities as inputs to the neural network.
REFERENCES:
patent: 4385285 (1983-05-01), Horst et al.
patent: 4724542 (1988-02-01), Williford
patent: 4993068 (1991-02-01), Piosenka et al.
patent: 5265008 (1993-11-01), Benton et al.
patent: 5355411 (1994-10-01), MacDonald
patent: 5434928 (1995-07-01), Wagner et al.
patent: 5497430 (1996-03-01), Sadovnik et al.
patent: 5559895 (1996-09-01), Lee et al.
patent: 5590218 (1996-12-01), Ornstein
patent: 5613012 (1997-03-01), Hoffman et al.
patent: 5615277 (1997-03-01), Hoffman
patent: 5621809 (1997-04-01), Bellegarda et al.
patent: 5668897 (1997-09-01), Stolfo
patent: 5679940 (1997-10-01), Templeton et al.
patent: 5748780 (1998-05-01), Stolfo
patent: 5748783 (1998-05-01), Rhoads
patent: 5754697 (1998-05-01), Fu et al.
patent: 5764789 (1998-06-01), Pare, Jr. et al.
patent: 5781654 (1998-07-01), Carney
patent: 5799311 (1998-08-01), Agrawal et al.
patent: 5802199 (1998-09-01), Pare, Jr. et al.
patent: 5819226 (1998-10-01), Gopinathan et al.
patent: 5898777 (1999-04-01), Tycksen, Jr. et al.
Therrien, “Decision Estimation and Classification”, 1989, pp. 30-35.*
Lippmann, “An Introduction to Computing with Neural Nets”, Apr. 1987, pp. 4-22, IEEE ASSP, vol. 4, #2.
Johnson Timothy M.
Sierra Patent Group Ltd.
Synaptics Incorporated
LandOfFree
Identity verification methods does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Identity verification methods, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Identity verification methods will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2927387