Ocado has unveiled a fraud detection service for its shoppers that uses machine learning (ML) techniques and Google Cloud to predict and defend against fraudulent transactions before they take place.

Customers’ order information is stored and analysed using BigQuery, a data analysis platform provided by Google Cloud, before being transferred to Cloud and Datastore storage using Dataflow. Stored data can be quickly and easily accessed by machine learning software in the Cloud to produce models of customer orders. Ocado’s bespoke software, powered by open-source machine learning platform TensorFlow, compares stored data to the Cloud’s models to make real-time predictions of customer shopping habits.

“A machine learning model … can learn and adapt far quicker, evolving based on the current environment and even predicting future trends,” Ocado said. “The model can also look at many more factors than a human or fixed rule based engine can.”

Frauds are expected in one out of every 1,000 orders, a rate of 0.1%, so Ocado has taken steps to ensure its fraud detection system is as effective as it can be. The company gave the system data on past orders, including fraudulent ones, to ensure it could make accurate predictions. It also hosted ‘hackdays’ where data scientists and software engineers attempted to hack the system to expose vulnerabilities and test new models.

The model has improved Ocado’s precision in detecting fraud ‘by a factor of 15 times’ and the company is now investigating algorithms that could explain the process behind the predictions in more detail. This would enable Ocado to share information between retailers, exposing the system to a broader range of data and improving its accuracy.

“The motivation behind using ML for fraud detection was twofold: speed and adaptability,” said the company.

“Machines are fundamentally more capable of quickly detecting patterns compared to humans. Also, as fraudsters change their tactics, machines can learn the new patterns much quicker.