Reduced manual review of handwritten checks
Models that use OCR and deep learning to scan checks, process data and verify signatures
Leverage historical database of previously scanned checks, including those known to be fraudulent
Models that assign risk scores to scanned checks, flagging them as good, fraudulent or needing further review.
Increased model accuracy of detection
Create Computer Vision model and build custom recipes that will generate features/variables that will provide a probabilistic estimation of the check being fraud or not.
Check stock validation (CSV) for counterfeit detection
Automated signature verification (ASV) for forgery detection
CAL/LAR discrepancy and payee name verification (PNV) for alteration detection
Thresholding rules, flagging major variations in check stock and signatures which may indicate a potential fraudulent payment
Once reviewed, feedback is provided back to the system to validate a “good” or “bad” payment.
H2O AI App: Check Fraud Detection, built with H2O Wave, SDK Kit for Data Scientists, can perform:
- Data Preparation (Bank data, customer data, customer images(signatures, etc))
Unsupervised Machine Learning, Neural Network, Predictive Modelling
Users can tune multiple parameters such as Sampling Ratio, number of hidden layers, epochs, etc.
Input data - cheque scan/photo
Dashboard containing the following major features:
Flag whether there is a major variation in payment made.
Flag whether operator review is needed.
Textual insights on explaining the prediction.
Probability of the cheque being fraudulent.
Automatic Signature Verification result, forged or not.
To interact with the Check Fraud Detection app, you must log in with your H2O Cloud Account. If you do not have an account, you can sign up for a 90-day free trial and create your account today. This app demonstrates the use of machine learning as part of an overall plan to minimize employee attrition. Users can view predictions of employee departure, forecast churn rates, and identify relevant factors contained in employee data.