An absence of high-quality data – the result of privacy trumping security – is one of the most serious impediments facing researchers as they attempt to use machine learning to tackle the twin problems of money laundering and human trafficking, says Mathab (Maria) Kamali, a data scientist with Thomson Reuters.
Speaking Wednesday at the Communitech Data Hub in Waterloo, Kamali told a lunchtime audience that low-quality data is crimping efforts to fully deploy assets like artificial intelligence to fight crime.
“If governments, banks and financial institutions work together while protecting the privacy of their customers and [then] share the data, it would give good-quality data to experts to work on it,” Kamali said.
Thomson Reuters, which maintains an innovation lab at Communitech (one of seven it has established around the world), is working in the background to help its financial customers combat money laundering, which Kamali said amounts to five per cent of global GDP, or US$2 trillion annually.
Money laundering, explained Kamali, “is a network of transactions,” one that often spans borders, involves many people, and creates complex “nodes” that machine learning is uniquely able to identify.
“So it’s a network analysis problem,” said Kamali.
But financial institutions, she said, typically are averse to sharing information with other financial institutions, do an incomplete job of collecting transaction information, and are further crimped by privacy laws, preventing scientists from gathering the necessary data and building the robust databases that could identify trends.
“Once privacy concerns are resolved and a cloud-based system established to share the data, then it’s possible to obtain a 360-degree view of the customer and it would be possible to work proactively against money laundering networks,” she said.
Kamali is from Iran and came to University of Waterloo after her undergraduate work to do a Ph.D. in systems design engineering. Her thesis dealt with the application of machine learning in optimizing weather models. She joined Thomson Reuters last November.
Her son – ”he’s more famous than me,” she says with a laugh – is Soroush Ghodsi, one of the founders of Slik.ai, and, at 15, was the youngest Canadian entrepreneur to join the Y Combinator accelerator.
The capability of AI and machine learning, said Kamali, is advancing rapidly. The challenge going forward is to work towards an evolution of privacy rules that protect individuals but allow the deployment of technology and advanced analytics.
Human trafficking and related money laundering activities, she said, are increasingly sophisticated, requiring an equally sophisticated technological response.
