Evolving Intelligent Systems - Concept, Applications and Opportunities for Security Systems of the Future
Tuesday 16 December 2008, 1140-1205
Lecture Theatre 1, Management School Building
Research into innovative computational intelligence methods to deal with data streams in real time will be presented by Plamen Angelov of Communication Systems.
By using fuzzy rule based systems to capture knowledge from the data streams by on-line learning of both their parameters and structure a series of powerful computational engines were pioneered at Lancaster - evolving clustering (eClustering), classifiers (eClass family), predictors (eTS family), controllers (eControl). They can be seen as fuzzy blends of locally valid Gaussian filters and also as self-developing neuro-fuzzy systems. They possess a high level of adpativity to unknown environments and have been applied to a range of practical problems:
a) intelligent sensors in oil refining (CEPSA Total) and chemical industry (Dow Chemical);
b) on-line machine health monitoring and prognostics (Ford);
c) autonomous systems for passive sense and avoid algorithm (BAE Systems);
d) landmark recognition and self-localisation of robots;
e) cyber security (hacker attacks and intruders detection, user behaviour modelling);
f) surveillance: object detection and tracking.
This approach possesses significant potential to be used in the security systems of the future for the following reasons:
a) evolving intelligent systems are convenient and rigorous tool for integration of expert knowledge and learning from data and experience; b) they can integrate the behavioural and psychological aspects of a security system and technological (engineering, mathematical, statistical);
c) they can deal with uncertainties and linguistic variables such as Anxiety, Fear, Hesitation which are hard to be quantified otherwise;
d) they tolerate imprecision. Interest to this original methodology for designing innovative in-flight security systems has been expressed by companies such as ULTRA and Thales.
Such research can be a building block in the new Centre on Behavioural Security Technologies (CBEST) that combines the efforts across the Faculty (led by the Psychology Department, Prof. T. Ormerod it also involves Communication Systems, Computing, and Engineering Departments) and is currently in its infancy.