No results found

Your search did not match any results.

We suggest you try the following to help find what you're looking for:

  • Check the spelling of your keyword search.
  • Use synonyms for the keyword you typed, for example, try “application” instead of “software.”
  • Try one of the popular searches shown below.
  • Start a new search.


Trending Questions

Financial Crime with Machine Learning

Map Out Financial Crime with Machine Learning

In an already stringent regulatory climate, banks and financial institutions face new hurdles around anti-money laundering (AML) and antiterrorist financing (ATF) compliance. Advanced analytical technologies come as a huge respite for banks to combat financial crime.

Unfortunately, firms have been slow to adopt innovative technologies such as artificial intelligence (AI) and machine learning (ML) in their anti-money laundering programs. There is limited understanding of AI and ML and how they fit in within financial compliance programs. In addition, regulators and compliance officers are often concerned that artificial Intelligence and machine learning are “black boxes” whose inner workings are not clearly understood.

However, the 2018 joint statement by major regulatory bodies in the United States, including FinCEN and FDIC, emphasize the use of technology to fight financial crime. Advanced analytics offers the possibility of not only driving down the cost of financial compliance but also of finally flushing financial criminals out of the system.

Artificial Intelligence for Anti-Money Laundering

The high false positive rate in transaction monitoring has driven up the cost of compliance for many institutions.

ML algorithms constantly learn from historical and current data and evolve to discover and model changing criminal behaviour patterns. Based on the model, data quality, regulatory requirements and risk tolerance of the organization, the red flags that are identified are either reported as a case for investigation or suppressed as a false positive. Information from valid cases are then fed back into the models, thereby being incorporated as part of learning into future models.

Organizations are thus able to reduce false positives by up to 30 percent. However, although machine learning models reduce the number of false positives, this is just half the battle won. Red flags are merely an indicator of money laundering or terrorist financing activity but still require further investigations on the pattern of transactions and related customer activities before it can be reported as a case.

Mapping Cases with Graphs

Considering the case as a graph opens a whole new avenue for detection. Once the data has been sourced, investigators can view all entities, customer, account, address, external entity and financial institution, and the various relationships in the form of an intuitive graph representation.

A graph database can help map the connections and make sense of the data. Graph technology naturally indexes data by relationships for detection and investigation and generates red flags in an entity network relating to all involved parties. The individual event scores are then consolidated to draw an overall case score.

Graph similarity relates similarities between two graphs. With graph similarity, the machine learning model is not limited to specific point in time risk indicators (such as transaction volume, transaction count, risk level, and number of parties) but expands to overall money movement pattern and the customer network. This is a more efficient way of comparing new versus previous cases leading to a better outcome.

Keeping up with Financial Criminals

Traditionally, machine-learning detection systems were designed to detect anomalies, not patterns. By adding the graph technology layer, machine learning transforms how we can detect the probability of suspicious activity in anti-money laundering and antiterrorist financing programs.

Graph analytics and graph similarity has numerous key applications in diverse fields (such as social networks, image processing, biological networks, chemical compounds, and computer vision). This makes it easier for anti-money laundering teams to access existing algorithms.

The challenge therefore is now how financial institutions will start keeping up with the financial criminals, but when.

Related Products and Solutions

Graph Analytics and Machine Learning

A portal into the financial crimes data lake. Apply machine learning, advanced analytics, and automation to turn raw data into actionable intelligence.

Transaction Monitoring

Next-generation AML compliance tools to analyze and mine suspicious patterns and transactions.

Detect, Investigate and Prevent Fraud

Combine real-time monitoring and correlation with sophisticated analytics and case management.

Improve Investigation Efficiency

Apply the latest advancements in machine learning to reduce false positives and prioritize alerts for investigation.