Admissions are open for starting in 2017. See below for various topics and PhD positions. Students are strongly advised to contact immediately and apply by early December, preferably sooner.
In an interconnected world, networks underlie all our social and technological systems. Examples of interconnected systems range from social networks and Internet to road networks and power grids. The increasing availability of data is now making it possible to analyse these networked systems at a high resolution and to look for patterns in their behaviour. Network science is the multidisciplinary subject spanning mathematics, informatics, physics and statistics that studies properties of networks and their relation to complex systems. See the STN course for more information.
Analysis of networks and dynamic processes. This topic is about study of emerging networks and data about networked processes. We are looking to build better analytic and learning techniques to identify hidden patterns in network data such as propagation of information in social networks or traffic in road networks.
Fast algorithms for network mining. Handling large networks requires efficient methods to compute their properties. This topic is about developing efficient algorithms that take advantage of modern systems with multiple parallel machines to speed up relevant computations.
Geometric techniques are fundamental to computing. In this topic, we are interested in applying geometry and topology for data mining. For example, spatial and sensor data can be handled better with these techniques. We are particularly interested in geometric information processing techniques.
Internet of things is emerging as the omnipresent computing platform. We already have mobile devices present almost everywhere, and now we are seeing sensors and small computers being embedded in everything from buildings to cars. These devices can produce large quantities of data that is challenging to process and store. Fast inference is critical in keeping up with the stream of data from millions of sensors. Our goal is to develop methods to handle data and analysis rapidly, on demand.
Data mining on mobile phones and sensors. Mobile devices and modern sensors are equipped with computers. Together, the devices wield enormous computational power. By utilising this power, we have the opportunity to collaboratively process data and learn patterns efficiently right at the source of the data — without creating separate servers for them, or having to upload and store all the data.
Fast data mining for IoT. On the server or cloud side, we face the challenge of handling information streaming in from many devices. In large quantities of data, much of the data can be redundant or noisy. The challenge we face is to quickly extract the essential data items and summary information. The focus here is on speed and summary computation, to enable more in depth analysis down the line.
See here for instructions.