Making theory practical
We are building the foundations for data driven decision making through machine learning. Our goal is to build upon the success of predictive models, using them as part of larger systems. In particular we want to understand how adaptive systems interact. As basic machine learning components such as binary classifiers become well understood and easy to use, the key challenge is to understand how the uncertainties of the machine learning component affects the surrounding system. In particular we wish to understand the interface itself. We do this in a bottom up approach by building prototypes that contain multiple machine learning components:
- Event detection and maintenance planning with an aircraft manufacturer
- Identification of potential disease pathways with collaborators in genomics and genetics
- Understanding the economics of online attention using social media data
- Choosing interesting phenomena from large astronomical surveys
- Scheduling and planning of urban traffic
Based on these examples we aim to find commonalities and patterns which will inform are more complete theory of machine learning systems.
For our research output, please have a look at each individual’s webpages.
The video above gives a short explaination of what is machine learning for the lay audience.