Secured transfer instruments
Disclosed are systems and methods for generating electronic instruments that implement digital transfers. The systems convert instruments to an electronic format using a digital imaging source that outputs image data. The image data is processed to determine content elements and segments of the electronic instrument and to extract transfer data. Neural networks implement artificial intelligence and machine learning technology that is used to secure the transfer instrument and detect inconsistencies or errors in the data.
1 . A system to secure electronic transfer instruments comprising a first computer including at least one processor and a memory device storing data and executable code that, when executed, causes the at least one processor to:
transmit system configuration data to a network computer that compares the system configuration data to stored system configuration data and returns a database entry and transfer activity;
activate a camera coupled with the computer, wherein the camera captures image data that comprises a transfer instrument image;
read and convert the image data to machine encoded content elements to identify text characters on the transfer instrument image;
using a neural network, segment the transfer instrument image into a plurality of components, wherein at least one of the components comprises transfer data, wherein the neural network executes a clustering analysis, wherein the neural network comprises one of a convolutional neural network (CNN), a Hopefield network, a Helmholtz Machine, a Kohonen Network, Sigmoid Net, a Self-Organizing Map, or a Centroid Neural Network;
extract the transfer data from the transfer instrument image, wherein the processor (i) reads machine encoded content elements from the component, and (ii) converts groups of machine encoded content elements to transfer data elements; and
secure the transfer instrument by using the transfer data elements, the database entry, and the transfer activity to determine a Secure Score that is compared to Secure Threshold to generate a pass indicator or a fail indicator, wherein:
when the pass indicator is generated, the transfer data is stored as transfer activity data, and
when the fail indicator is generated, the transfer data is not stored as transfer activity data.
2 . The system of claim 1 , wherein the neural network is used to secure the transfer instrument.
3 . The system of claim 2 , wherein the neural network comprises the CNN.
4 . The system of claim 2 , wherein the CNN comprises at least three layers.
5 . The system of claim 1 , wherein, prior to causing the processor to create the electronic transfer instrument, the transfer instrument image is enhanced using one or a combination of binarization, de-skewing, de-warping, or de-speckling the transfer instrument image.
6 . The system of claim 1 , wherein the executable code, when executed, causes the at least one processor to:
utilize the transfer data, client data, and transfer activity data to generate known labeling data;
generate an error rate by comparing the pass indicator or the fail indicator to the known labeling data; and
train the neural network by adjusting one or more neural network weighting coefficients to reduce the error rate.
7 . A system to secure electronic transfer instruments comprising a first computer including at least one processor and a memory device storing data and executable code that, when executed, causes the at least one processor to:
load transfer activity data;
load image data generated by a camera coupled with a client computer, wherein the image data comprises a transfer instrument image;
read and convert the image data to machine encoded content elements to identify text characters on the transfer instrument image;
convert groups of machine encoded content elements to transfer elements;
using a neural network, secure the transfer instrument by using the transfer elements and the transfer activity data to determine a Secure Score that is compared to Secure Threshold that results in a transfer pass indicator or a transfer fail indicator that is transmitted to the client computer;
utilize transfer data, client data, and the transfer activity data to generate known labeling data;
generate an error rate by comparing the transfer pass indicator or the transfer fail indicator to the known labeling data; and
train the neural network by adjusting one or more neural network weighting coefficients to reduce the error rate.
8 . The system of claim 7 , wherein the neural network comprises an architecture selected from one of: (i) a long short term memory; (ii) a recurrent networks; (iii) an Elman recurrent network; (iv) a convolutional neural network (CNN); (v) a multilayer perceptron network; (vi) a TensorFlow network; (vii) a MxNet networks; (viii) a PyTorch network; (ix) a Keras network; or (x) a Gluon network.
9 . The system of claim 7 , wherein: the transfer data comprises transfer value data, and the Secure Threshold increases when the transfer value data increases.
10 . The system of claim 7 further comprising a Deposit Service Router and Parameters and Threshold Database, wherein:
the Parameters and Threshold Database comprises an active path threshold; and
the Deposit Service Router passes the transfer instrument through an Active Path when the active path threshold is met or an Onboard Path when the active path threshold is not met.
11 . The system of claim 7 , wherein: a second neural network is used to read and convert the image data to machine encoded content elements, wherein the second neural network comprises a network architecture selected from one of a convolutional neural network (CNN), a Hopefield network, a Helmholtz Machine, a Kohonen Network, Sigmoid Net, a Self-Organizing Map, or a Centroid Neural Network.
12 . The system to secure electronic transfer instruments of claim 7 , wherein, prior to causing the processor to create the electronic transfer instrument, the transfer instrument image is enhanced using one or a combination of binarization, de-skewing, de-warping, or de-speckling the transfer instrument image.
13 . A system to secure electronic transfer instruments comprising a first computer including at least one processor and a memory device storing data and executable code that, when executed, causes the at least one processor to:
load transfer activity data;
load image data generated by a camera coupled with a client computer, wherein the image data comprises a transfer instrument image;
read and convert the image data to machine encoded content elements to identify text characters on the transfer instrument image;
using a neural network, segment the transfer instrument image into a plurality of components, wherein at least one of the components comprises transfer data, wherein the neural network executes a clustering analysis, wherein the neural network comprises one of a convolutional neural network (CNN), a Hopefield network, a Helmholtz Machine, a Kohonen Network, Sigmoid Net, a Self-Organizing Map, or a Centroid Neural Network;
extract the transfer data from the transfer instrument image, wherein the processor (i) reads machine encoded content elements from the component, and (ii) converts groups of machine encoded content elements to a transfer element; and
secure the transfer instrument by using the transfer element and the transfer activity data to detect transfer tags, wherein: (i) when the transfer tags are detected, the transfer data is sent to a Deposit Platform, and (ii) when the transfer tags are not detected, the transfer data is not sent to the Deposit Platform.
14 . The system of claim 13 , wherein
the neural network is used to secure the transfer instrument.
15 . The system of claim 13 further comprising a Deposit Service Router and Parameters and Threshold Database, wherein:
the Parameters and Threshold Database comprises an active path threshold; and
the Deposit Service Router passes the transfer instrument through an Active Path when the active path threshold is met or an Onboard Path when the active path threshold is not met.
16 . The system of claim 13 , wherein the executable code, when executed, causes the at least one processor to:
utilize the transfer data, client data, and transfer activity data to generate known labeling data;
determine an error rate of the neural network based at least in part on the known labeling data; and
train the neural network by adjusting one or more neural network weighting coefficients to reduce the error rate.