Anti Money Laundering (AML) Pattern Detection using deep learning

Traditional AML pattern detectors have relied on rules. Money laundering and terrorist financing cases can be extremely complex, often involving many players implicated in transnational and covert illicit activity. The investigations are often time and resource intensive. An example of a sample dataset for AML might be

  • type - CASH-IN, CASH-OUT, DEBIT, PAYMENT and TRANSFER.
  • amount - amount of the transaction in local currency.
  • nameOrig - customer who started the transaction
  • oldbalanceOrg - initial balance before the transaction
  • newbalanceOrig - new balance after the transaction
  • nameDest - customer who is the recipient of the transaction
  • oldbalanceDest - initial balance recipient before the transaction. Note that there is not information for customers that start with M (Merchants).
  • newbalanceDest - new balance recipient after the transaction. Note that there is not information for customers that start with M (Merchants).
  • isFraud - This is the transactions made by the fraudulent agents inside the simulation. In this specific dataset the fraudulent behavior of the agents aims to profit by taking control or customers accounts and try to empty the funds by transferring to another account and then cashing out of the system.
  • isFlaggedFraud - The business model aims to control massive transfers from one account to another and flags illegal attempts. An illegal attempt in this dataset is an attempt to transfer more than 200.000 in a single transaction.

What does your dataset look like? Please note that this dataset is very basic. It does not contain information about the Customer. Which mobile device did they use. Their address, etc. However, even with this dataset Sublime AI's deep learning system can start to learn some rules especially using LSTM networks. The system is given a sequence of these transactions by a given customer and the task is then to predict if the transaction is fraud. Depending on your dataset the deep learning system can automatically detect rules such as :

  • unverfied documents
  • multiple tax ID numbers
  • sparse business information
  • unexpected change in currency transaction number type or volume
  • transaction patterns are different from other similar businesses, etc.

If you want to see how deep learning can find new AML patterns and significantly increase the accuracy please get in touch with a format of the AML dataset you posses in your company

Fraud Detection

Deep learning algorithms can quickly sift through and analyze vast quantities of data and uncover transactional anomalies or suspicious patterns. Unsupervised Deep Learning can be used to detect anamolies. An example of a sample dataset is

  • type - CASH-IN, CASH-OUT, DEBIT, PAYMENT and TRANSFER.
  • amount - amount of the transaction in local currency.
  • nameOrig - customer who started the transaction
  • oldbalanceOrg - initial balance before the transaction
  • newbalanceOrig - new balance after the transaction
  • nameDest - customer who is the recipient of the transaction
  • oldbalanceDest - initial balance recipient before the transaction. Note that there is not information for customers that start with M (Merchants).
  • newbalanceDest - new balance recipient after the transaction. Note that there is not information for customers that start with M (Merchants).
  • isFraud - This is the transactions made by the fraudulent agents inside the simulation. In this specific dataset the fraudulent behavior of the agents aims to profit by taking control or customers accounts and try to empty the funds by transferring to another account and then cashing out of the system.
  • isFlaggedFraud - The business model aims to control massive transfers from one account to another and flags illegal attempts. An illegal attempt in this dataset is an attempt to transfer more than 200.000 in a single transaction.

What does your dataset look like? Please note that this dataset is very basic. It does not contain information about the Customer. Which mobile device did they use. Their address, etc. However, even with this dataset Sublime AI's deep learning system can start to learn some rules especially using LSTM networks. The system is given a sequence of these transactions by a given customer and the task is then to predict if the transaction is fraud. Depending on your dataset the deep learning system can automatically detect rules such as :

  • unverfied documents
  • multiple tax ID numbers
  • sparse business information
  • unexpected change in currency transaction number type or volume
  • transaction patterns are different from other similar businesses, etc.

If you want to see how deep learning can find new AML patterns and significantly increase the accuracy please get in touch with a format of the AML dataset you posses in your company