Fraud is a big deal -- it costs the economy several billion dollars a year. For many years, machine learning has been the tool of choice for solving fraud. Unfortunately, machine learning is only useful when you have lots of high-quality labeled data. In the real world it's difficult, expensive, or impossible to get this high-quality labeled data.
In my talk I'm going to look at the problem of fraud at a high level, how it can be solved using unsupervised techniques that don't require labels, and the infrastructure required to implement it.