Fighting money laundering and fraud in financial services through AI

Identifying patterns of transactions, behaviors and anomalies

Electronic transaction volumes are growing exponentially within financial systems, and the ability to identify both unusual and usual patterns of behavior is critical to monitoring the safe and efficient functioning of payment systems. Data analysis is key in identifying behaviors and patterns to detect fraud and money laundering, but with financial services organizations encountering unparalleled levels of complexity with their data, intelligent solutions are needed to:

•Analyze relationships and connections between data
•Uncover adaptive and predictive insights
•Detect the shape of connections between data
•Explain the factor and reason of the result

Currently, most AI solutions analyze tabular data stored in relational databases (RDBs) which provide only a limited capability to detect: relationships between data, hidden relationships, or to grasp key points between complex data. In the fight against fraud and money laundering, a new approach is needed, one that is capable of analyzing complex relationships. This approach is graph AI.

Graph AI is particularly adept at analyzing complex data relationships, detecting trends and relationships that would go unnoticed by RDB-based solutions. It detects any hidden data relationships, key points between complex data, and clearly explains the results and reasons for the predictions made. And so, when applied to anti-money laundering and fraud, graph AI is capable of looking into whether two supposedly different individuals can actually be the same person.

“Fujitsu Finplex EnsemBiz powered by Deep Tensor can analyze relationships between data to provide high accuracy predictions, discover trends that were previously undiscoverable and detect circular relationships to help deliver a sustainable future.” Daniel Ratiu

Eliciting new insight from graph data for a sustainable future within financial services

Capable of detecting and understanding highly complex relationships, Fujitsu Finplex EnsemBiz powered by Deep Tensor AI enhances compliance capabilities and combats fraud with a graph-based data science approach. It analyzes relationships within the data and provides convincing predictions to help solve underlying social issues.
Unlike traditional RDBs, graph structure represents diverse and complex data in the form of a network somehow similar to human brain connections. Deep Tensor analyzes and utilizes relationships between graph structure data automatically and efficiently, making it possible to detect illegal transactions such as circular trading fraud which has proven difficult to identify with conventional technologies. It uses advanced analysis and machine learning to derive new knowledge from graph structure data representing connections between people and goods. By converting data into a graph model, it’s easier to:

•Obtain convincing predictions with clear relationships between data - ‘explainability’ of AI
•Discover common trends that were previously unnoticed allowing for new business insights and opportunities

Deep Tensor eases AI concerns by making it possible to understand and verify the reasoning behind the outputs of a system’s decision-making processes, therefore eliminating the ‘black box’-like perception of AI. When compared to other graph AI technologies such as GNN-Explainer, it has 20% higher prediction accuracy (e.g. 99% compared to 44.1%) and 30% higher explanation accuracy (e.g. 78.3% compared to 15.9%). Furthermore, it can also detect recurrent relationships, which GNN-Explainer cannot. Used within the financial services industry it can successfully analyze:

•Transaction history to detect financial transaction fraud
•Relationships and connections between high-risk companies and people to detect money laundering
•Business similarities to help business matching

Using data connections and data analysis to realize a better world

The benefits of leveraging EnsemBiz are not limited to financial services with applications such as anti-money laundering, risk assessment, fraud detection and streamlining loan screening processes. The aim is to realize a sustainable world by setting ambitious goals and constantly tracking progress towards them. For example, future applications may include helping to meet CO2 reduction targets by providing a visual and easy to grasp means to detect, track and understand data which can then be used to help both businesses and individuals lower their carbon footprint and emissions. Using this new technology solution, we hope to create a world where people can live safely, without worry, by finding connections between people and things in all settings of their lives whether that’s within insurance, healthcare, law enforcement, pharmaceuticals, or security services – such as:

•Chemical compounds for medicine and drug discovery
•Tracking viruses in healthcare
•Social media for wholesale and retail
•Communication logs for security

Contact us to find out more about Finplex EnsemBiz powered by Deep Tensor and the benefits it can deliver whether that’s detecting fraudulent transactions or helping you to move towards a more sustainable business future.