Fighting crime with graph learning

Organized crime inflicts human suffering on a massive scale: the Mexican drug cartels have murdered 150,000 people since 2006; upward of 700,000 people per year are “exported” in a human-trafficking industry enslaving an estimated 40 million people. These nefarious industries rely on sophisticated money-laundering schemes to operate.

Despite tremendous resources dedicated to anti-money laundering (AML), only a tiny fraction of illicit activity is prevented. The research community can help. Mark Weber (MIT-IBM Watson AI Lab) explores how to map the structural and behavioral dynamics driving the technical challenge, and he reviews AML methods both current and emergent. You’ll get a first look at scalable graph convolutional neural networks for forensic analysis of financial data, which is massive, dense, and dynamic. Mark outlines preliminary experimental results using a large synthetic graph (1M nodes, 9M edges) generated by a data simulator called AMLSim, and he considers opportunities for high performance efficiency, in terms of computation and memory, and shares results from a simple graph compression experiment, all of which supports the working hypothesis that graph deep learning for AML bears great promise in the fight against criminal financial activity.

Prerequisite knowledge

• A basic understanding of data science and graph structures
• Experience with finance (useful but not required)

What you'll learn

• See why graph deep learning is a powerful tool for finance and other applications

This session is from the 2019 O'Reilly Artificial Intelligence Conference in San Jose, CA.