ML for fraud detection

Machine-learning technique employed to reduce false positives in fraud detection



Consumers’ credit cards are declined surprisingly often in legitimate transactions. One cause is that fraud-detecting technologies used by a consumer’s bank have incorrectly flagged the sale as suspicious. MIT LIDS researchers developed a new machine-learning technique to drastically reduce these false positives, saving banks money and easing customer frustration.

Tested on a dataset of 1.8 million transactions from a large bank, the model reduced false positive predictions by 54 percent over traditional models, which the researchers estimate could have saved the bank 190,000 euros (around $220,000) in lost revenue.

The joint research team was composed by three LIDS researchers (led by LIDS principal investigator, Dr. Kalyan Veeramachaneni) and two researchers from BBVA (Banco Bilbao Vizcaya Argentaria), Madrid, Spain. They used the Deep Feature Synthesis algorithm to automatically derive behavioral features based on the historical data of the card associated with a transaction. It generated 237 features (>100 behavioral patterns) for each transaction and used a random forest to learn a classifier.

ml2

Two researchers from BBVA, Madrid, Spain, were collaborators for this project.

The technology is detailed in a paper entitled with “Solving the false positives problem in fraud prediction using automated feature engineering”. It was presented in the European Conference on Machine Learning and Principles and Practices of Knowledge Discovery in Databases (ECML-PKDD), Dublin, Ireland, September 10-14, 2018.

ml3

The project was led by a LIDS Principal Investigator, Dr. Kalyan Veeramachaneni.