Description: Smart applications anticipate typical human behavior. A key to effectively predict human behaviour and act proactively is the ability to create models of individual or group behaviour from a variety of data obtained from sensors and other low-level events.
While user profiles are being used extensively in ubiquitous computing applications, there are various shortcomings and limitations related to the expressiveness of the models, and to the dynamic construction and updating of the models. These models also lack support for context-dependent inference and prediction of most-likely future behaviour of a single individual or a group as a whole.
In this PhD project, the candidate will investigate techniques to model, mine, and learn routines in order to recognize human behavior and activities within a particular context or situation. The methodology shall be based on event-based evidence of interactions with smart systems and state-of-the-art complex event processing techniques to identify relevant patterns.
The goal of this research is to be able to predict and proactively anticipate the most likely future behaviour of an individual or a group as a whole. The research will be carried out in the area of e-health applications and ambient assistent living.
Applicants should have a MSc in computer science or similar and should be highly motivated to work in the field of software engineering and be familiar with the principles of event-based systems. Basic knowledge of middleware for distributed systems and AI algorithms for prediction and learning is strongly encouraged.
Key words: middleware, ubiquitous computing, distributed systems, event-based systems
Latest application date: 2011-10-03
Financing: available
Type of Position: scholarship
Duration of the Project : 4 years
Research group: Department of Computer Science
Apply to Click here to apply to this project
No comments:
Post a Comment