Extracting knowledge from data
Luckily several external factors have aligned to provide a solution to many of these issues.
- The increase in data availability, dubbed ‘Big Data’.
- Improvements in processing power delivered by GPUs.
- Cloud access.
- The rise of Machine Learning techniques to detect unknown or unclear, yet significant, patterns in data.
- The acceptance in many sectors that Machine Learning can add much needed insight and automation.
Taken together, these factors promote a sea-change in the process of detecting fraud and preventing money laundering.
TruNarrative delivers advanced detection based on decision tree and proven Machine Learning techniques. Our clients, the consortium and external sources provide the data. The TruNarrative platform continually analyses masses of this data, seeking patterns which reveal knowledge that otherwise would be overlooked. This enables us to automate much of the analytics and improves detection, whilst minimising false positives across the entire fraud and AML process.
Perfect balance of machine & human collaboration
Machine Learning and human insight combine in the most advanced solution. Human judgement contributes throughout.
We use this process to build our Machine Learning models. These models are then refined automatically as new data enters the TruNarrative platform. The platform gets smarter on every new event and learns 24×7.
We also constantly review all decision tree rules held in our service against your performance data. This ensures you are not deploying inefficient or ineffective rules. We flag which rules you deploy that could be more effective or we find new rules not considered before. You remain the expert.
Not a Black Box
The idea that Machine Learning is a “magic black box” makes people wary. Black box solutions provide little or no explanation of why a specific customer has been flagged as suspicious. We understand the concern.
So TruNarrative uses a selection of “white box” Machine Learning methods. These show you the reasoning for any customer flag. Of course we can also deploy neural networks at single and multi-layer should you require or even champion them against each other.
You stay in control. As you become familiar and comfortable with the benefits and accuracy of our Machine Learning you can selectively allow it to make more automated decisions. However much or little you choose to automate, you can always stay in full control of the final decision process.