US 11,989,702 B2
Automated validation of digit sequences in transactions
Ido Meir Mintz, Tel Aviv (IL); Alexander Zhicharevich, Hod Hasharon (IL); Shlomi Medalion, Lod (IL); and Tom Jacobe, Holon (IL)
Assigned to Intuit Inc., Mountain View, CA (US)
Filed by Intuit Inc., Mountain View, CA (US)
Filed on Dec. 12, 2019, as Appl. No. 16/712,772.
Prior Publication US 2021/0182803 A1, Jun. 17, 2021
Int. Cl. G06Q 20/04 (2012.01); G06F 17/16 (2006.01); G06N 3/02 (2006.01); G06N 20/00 (2019.01)
CPC G06Q 20/042 (2013.01) [G06F 17/16 (2013.01); G06N 3/02 (2013.01); G06N 20/00 (2019.01)] 18 Claims
OG exemplary drawing
 
14. A method, comprising:
extracting, from a transaction, one or more categorical features and a digit sequence;
embedding the one or more categorical features in a categorical vector;
embedding the digit sequence in a digit sequence matrix;
encoding, using a first neural network layer of a trained machine learning model, a combination of the categorical vector and the digit sequence matrix to obtain a first output;
encoding, using a second neural network layer of the trained machine learning model, a combination of the categorical vector and the first output to obtain a second output; and
classifying the digit sequence as invalid, the classifying comprising applying the trained machine learning model to the second output, wherein applying the trained machine learning model reinforces attention on a dependency between the one or more categorical features and the digit sequence.