Many revenue assurance and fraud management solutions on the market focus on identifying a list of unusual patterns to generate an alarm for follow up. In this lab, WeDo Technologies is developing risk management models that when trained with a bias towards identifying contextualized unusual patterns, are more likely to exclude the 'normal' discrepancies, in order to reduce false positives. Using machine learning, we can draw relationships between different elements of data and provide augmented contextual awareness, to provide more accurate risk management alerts.
In other words, the models are trained to recognize unusual non-contextualized and unusual contextualized patterns in order to generate alarms only for the non-contextualized.
