Ocado is using machine learning to identify fraud in its systems.
The online supermarket’s technology division, Ocado Technology, says the artificial intelligence technology has improved fraud detection on the Ocado Smart Platform by a factor of 15.
The company wanted to find a way of predicting and recognising potential fraud incidents, and chose machine learning because of its speed and adaptability.
“Machines are fundamentally more capable of quickly detecting patterns compared to humans,” said Ocado’s Holly Godwin. “Also, as fraudsters change their tactics, machines can learn the new patterns much quicker.”
The new system is based on data from past orders, including cases of fraud. From this, the developers created a list of features which included the number of past deliveries, the cost of baskets, and other information.
They then created a machine learning algorithm, combined with open-source software library Tensorflow and running in Google Cloud.
As well as learning and adapting more quickly than other solutions, a machine learning model can look at a multitude of factors, Godwin explained.
“The work of fraud agents is then made more manageable, as they no longer have to frantically analyse thousands of data points to establish fraud,” she said.
“Instead, they simply perform a final check to confirm whether they should cancel the order or not based on the prediction made by the model; it’s a perfect case of humans and machines working together in harmony.”
Next, the developers are investigating algorithms that could allow the company to explain its predictions in more detail, assessing whether learnings can be transferred from one retailer to another, and considering what tools could help to streamline the process.