PhD - Analysis of heterogeneous data

Lancaster University

Lancaster, UK

 

The research is funded by SELEX Gallileo, UK (www.selexgalileo.com ) and will be conducted in joint collaboration with the industrial sponsor. The project is part of a Centre of Excellence comprising SELEX Galileo and 5 UK universities. The aim is of this studentship is to address both theoretical and application in extracting knowledge from a large amount of heterogeneous data, e.g. optical, infrared, Synthetic Aperture Radar images and burst illumination laser. The main focus will be on patterns matching and statistical data mining methods. 

 

Methodology:

 

Bayesian distributed data fusion methods and probabilistic graphical models (such as Bayesian networks, tree types of models), nonparametric inference methods will be studied. One approach that will be investigated is based on hierarchical Dirichlet models, and the impact of the prior on the decision making process, nonparametric models for inference and fusion. Nonparametric methods are a class of methods that allow the data to determine the complexity of the model. This prior knowledge will afford to account for prior information in an efficient way. Data of various types will be considered, such as from the airborne sensors, and ground sensors. The reliability metrics will include (but are not limited to) accuracy, computational time and minimum communications. Performance studies will be conducted between centralised, distributed and hybrid architectures. Some of the challenges that will be considered are: fusion of high dimensional data, both on-line and off-line, asynchronous arrival of the data, detection of faults and abilities to cope with missing or rare data, followed by an inference. In case of faults the sensor network should be reconfigured in a way to cope with the missing data and faults. Then the information extracted from the data is meant to be provided to the Unmanned Aerial Vehicles or operators for achieving better situation awareness. The results from the fusion algorithms will be used for recognition and classification of stealthy objects. 

 

Areas of expertise/ qualifications:

 

Statistical and probabilistic methods, numerical methods, digital signal processing, optimisation. Candidates with solid mathematical background are invited to apply for this position. Applicants should hold (or expect to obtain) a minimum upper-second class honours degree or equivalent in a discipline related with electrical engineering, aerospace engineering, statistical science or computer science. Willingness to study across scientific disciplines and a willingness to learn fast new areas of research will also be essential. Because this studentship is a part of a project and includes communication with partners of a large consortium, excellent communications skills are required. 

 

To apply please complete a PhD application form available at

(http://www.lancs.ac.uk/users/admissions/postgrad/pgform1.htm) and indicate on the form that you wish to be considered for the PhD studentship sponsored by Selex Gallileo, UK. The following documents are needed: a CV, a cover letter (including a motivation statement), a research proposal, and contact details of two referees. Further information can be obtained from Dr Mila Mihaylova (mila.mihaylova@lancaster.ac.uk). The PhD student will be based in Lancaster University, Infolab21 – a Centre of Excellence in ICT at the School of Computing and Communication Systems (http://www.scc.lancs.ac.uk/).