Graph Deep Learning
Paper: “Scalable Graph Learning for Anti-Money Laundering: A First Look” (presented at NeurIPS 2018)
Organized crime inflicts human suffering on a genocidal scale: the Mexican drug cartels have murdered 150,000 people since 2006; upwards 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. In this brief paper, we map the structural and behavioral dynamics driving the technical challenge. We review AML methods, current and emergent. We provide a first look at scalable graph convolutional neural networks for forensic analysis of financial data, which is massive, dense, and dynamic. We report preliminary experimental results using a large synthetic graph (1M nodes, 9M edges) generated by a data simulator we created called AMLSim. We consider opportunities for high performance efficiency, in terms of computation and memory, and we share results from a simple graph compression experiment. Our results support our working hypothesis that graph deep learning for AML bears great promise in the fight against criminal financial activity.
Graph Convolutional Networks (GCN) are opening up new possibilities for network analysis
We've seen deep learning do remarkable things on Euclidean data - audio, images, video. Not so much yet on graph data, until very recently. Graph data is structurally different; it's all about relationships between data. Think social networks, gene expression networks, knowledge graphs, you name it - graphs are all around us. In finance we can think about trading, hedging, and asset management, supply chain finance and optimization, lending and securitization. Each of these can use graphs to capture relationships and interactions between different types of entities, often with a time series component, and often in a dynamic setting. The problem is, deep learning on graph data is extremely difficult computationally due to the combinatorial complexity and nonlinearity inherent to graphs of any meaningful size and density. And it's precisely the information hidden in that complexity that makes graph data so interesting and important.
Recently we've seen a rapid and exciting acceleration of work on graph convolutional networks, or GCN's, with special attention to the question of scaling (see Kipf and Welling). With GCNs, we begin with certain attributes to describe the nodes and edges, and we use convolutions over the graph to pull out the hidden properties and patterns. This is called node embedding and the objective is to achieve a better vector representation of each entity. In laymens’ terms, you can think of each node asking the age-old question, “Who am I?” It’s really an existential question with infinite complexity, but we need a vector of finite length. So we have to bound the model or it’s going to take a prohibitively long time, and find a way do so without sacrificing accuracy. This is the challenge of scalability.
Earlier this year at ICLR, my colleagues Jie Chen and Tengfei Ma presented a new method called FastGCN. And if I may sing their praises for a moment, this work was a big step forward on scalability. FastGCN was able to beat previous speed benchmarks by two orders of magnitude. It does so by using a variant method for importance sampling and by performing integral transformations in node embedding to account for node inter-dependency.
Building on FastGCN, we're now exploring how we can further advance graph deep learning, and finance presents some interesting use cases. I’ve mentioned some of these, but one I find especially important is the challenge of anti-money laundering.
At a high level, money laundering is simply the concealment of criminal money flows via layered transfers involving multiple banks and/or legal businesses. Anti-money laundering (AML) is a regulatory requirement with five key components. Global costs of AML compliance run in the tens of billions of dollars and have been growing at about 15% per year since 2004. Unfortunately even with all that spending, the bad guys are still winning. Banks deal with extremely high false positive rates and Europol estimates that only 1% of criminal funds are confiscated. Penalties for AML non-compliance are brutal. For example, this summer the Commonwealth Bank of Australia was fined some $534 million for violations involving bugs in its "Intelligent Deposit Machines," which ostensibly used some sort of AI. A cautionary tale that calling something "intelligent" does not make it so.
But the social impact is even more profound. These criminal industries are causing human suffering on a genocidal scale. We have a $40 billion dollar human trafficking industry moving money through our financial system. 800,000 people are "exported" annually in a slave trade oppressing 40 million people.
Let that sink in for a moment.
This isn't some mundane matter of regulatory compliance. This is a matter of human dignity, and the research community can help.