Machine learning models that are hard to explain and what we can do about it

We are delighted to announce a talk by Dean Allsopp, titled Machine learning models that are hard to explain and what we can do about it.

Some popular and powerful machine learning models seem to be opaque, which means that we cannot straightforwardly assess them to understand what their behaviour will be or to gain insight into the domain which is modelled. This applies to both sub-symbolic models, such as neural networks, and also to complex symbolic models, such as decision tree ensembles. This talk will discuss why advanced models tend to be hard to explain, and techniques for analysing and understanding aspects of their behaviour.

Dean Allsopp has been seeking to extract information and value from data for over twenty years. After more than a decade of SQL programming, he transitioned to an analytics driven role to improve operational effectiveness. Currently freelance, he is exploring the potential of multiple database types and machine learning techniques to meet real business objectives.