Computational Intelligence (CI) is an emerging area of fundamental and applied research in the theory and design of intelligent systems. The particular form of CI research carried out in this laboratory includes neural networks, fuzzy sets, rough sets, a recent generalization of rough sets called near sets, evolutionary computation, machine learning, image processing and object recognition.
Evolutionary Computation are biologically-inspired methodologies aimed at global optimization. Genetic algorithms have been used by members of this group in solving the problem of matching 2D image segments.
Fuzzy sets (introduced by Lotfi Zadeh during the mid-1960s) provide a means of representing and processing linguistic or, in general, non-numeric information. They support a diversity of knowledge representation mechanisms. Fuzzy sets exploit imprecision in an attempt to make system complexity manageable. Hybrid fuzzy-rough neural networks have been used by members of this group to classify power system faults.
Image Analysis and Object Recognition This is a very active CI Lab research area. For example, both rough sets and near sets are used in image segmentation, image classification, measuring the similarity of images, and image quality estimation (see list of CI Laboratory Publications).
Machine Learning In recent years, considerable work has been done by this research group on theory and application of biologically-inspired adaptive learning. A bioproduct of this research has been the design of adaptive telegaming systems and a target-tracking system that learns. Learning is carried out in an episodic, ethological, approximation space-based framework (see list of CI Laboratory Publications).
Near sets are disjoint sets that are resemble each other. One set X is near another set Y to the extent that the description of at least one of the objects in X is similar to the description of at least one of the objects in Y. The hallmark of near set theory is object description and the classification of objects by means of features. Research in near sets includes approach spaces, definable sets, merotopic spaces, metric spaces, nearness spaces, object description, proximity spaces, tolerance spaces, and topology. The near set approach has proven useful in AI applications, data mining, feature selection, digital image (picture) analysis, content-based image retrieval, pattern recognition, target tracking and machine learning.
Neural networks offer powerful and distributed computing architectures equipped with significant learning abilities. They help represent highly nonlinear and multivariable relationships. They have already been successfully used in many system modeling and process control applications. During the past 8 years, members of this research group have introduced various forms of rough neural networks have introduced and applied in classifying power system faults.
Rough sets (introduced by Zdzislaw Pawlak during the early 1980s) result from approximation of one set by another set. Rough set theory has two main parts: knowledge description and set approximation. The hallmark of rough set theory is information granulation, especially if one considers granulation within the framework of an approximation space (see list of CI Laboratory Publications).
The CI laboratory engages in fundamental as well as applied research in Computational Intelligence. In parallel to ongoing research projects, the objective of the Laboratory is to provide the community with fully updated information on CI. It is an indispensable source of current information encompassing internal reports, relevant publications and providing with a number of links to selected sites worldwide.