In many applications, we not only care if a prediction is right, but we also want to understand how the model reached a certain conclusion.
For example, automatic fraud detection is often supervised by human analysts and the models produced from data need to be understandable and verifiable.

Machine learning has come to prominence in the last few years. But despite the progress made, the problem of learning automatically transparent structured models such as decision trees or rules is still particularly challenging.

In this talk we explore some machine learning techniques applied to structured data, giving an overview of the algorithms, the challenges involved, and the research landscape.