Instructor: Robert Kozma (202 Dunn, rkozma@memphis.edu )

A dynamical systems perspective on intelligence in computational systems is introduced. The methods of analysis are strongly biologically motivated, including hybrid fuzzy-neuro systems, emergence and chaos computing. Computational intelligence deals with the way these components are used for the automatic generation of knowledge in computational systems. In addition to computer science students, the course will be of interest to students in cognitive science as computational intelligence is closely related to modeling human cognitive behavior. Due to the interdisciplinary nature of the subject, we can expect interest also from various disciplines which apply the ideas of computational intelligence and soft computing to solve problems of adaptive control, robotics, intelligent human computer interaction, business/finance, among others.

The course gives an introduction to various aspects of modeling and transformation of information and knowledge in computers. It emphasizes that a dynamical systems perspective on intelligence is a key to understand intelligent behavior in biological and computational systems. Applications include neuro-fuzzy control, knowledge acquisition in large databases, financial data analysis, and biomedical fields. Implementations provided in MATLAB environment. PREREQUISITES: 6002 or 6001 or permission of the instructor.

This course requires high level of involvement and creative, research-oriented activities from each enrolled student. Grading will include the following components: class participation (15%), presentations on specific topics in class (15%), laboratory work (20%), project (50%). Students will develop a comprehensive project in their selected topic, including theoretical foundations, implementation, and testing using actual data environment.

August 28 Orientation

August 30 Classical artificial intelligence and connectionism

September 4 Introduction to soft computing principles: fuzzy-, neural-,

evolutionary-, genetic-, and chaos computing

September 6 Fuzzy set theory

September 11 Fuzzy relations

September 13 Fuzzy numbers

September 18 Linguistic descriptions and computational approaches

September 20 Fuzzy classifiers and fuzzy clustering

September 25 Fuzzy control

September 27 Implementation of fuzzy control

October 2 Foundations of dynamical systems theory and chaos

October 4 Quantitative characterization of chaos processes

Fall Break

October 11 Project outline discussion

October 16 Spatio-temporal chaos

October 18 Stochastic resonance, chaotic resonance

October 23 Chaotic self-organization and structure emergence in intelligent systems

October 25 Fundamentals of neural networks

October 30 Implementation of neural networks, neurocontrol

November 1 Fuzzy methods in neural networks

November 6 Neural methods in fuzzy systems

November 8 Hybrid neuro-fuzzy systems for knowledge monitoring

November 13 Structural learning and rule extraction from neuro-fuzzy systems

November 15 Optimization of neuro-fuzzy systems with genetic methods

November 20 Project presentations

Thanksgiving

November 27 Project presentations

November 29 Project presentations

December 4 Conclusions

Textbook (covering large part of the course but not everything):

• Fuzzy and Neural Approaches in Engineering, L.H. Tsoukalas, R.E. Uhrig, John Wiley & Sons, N.Y., 1997.

• Neuro-Fuzzy and Soft Computing: A Computational Approach to Learning and Machine Intelligence, J.S.R. Jang, C.T. Sun, E. Mizutani, Prentice Hall, 1996.
• Neuro-Fuzzy Pattern Recognition: Methods in Soft Computing, S.K.Pal, S.Mitra, Wiley, 1999.
• Neuro-Fuzzy Techniques for Intelligent Information Systems, N.K. Kasabov, R. Kozma, Springer, 1999.
• Nonlinear Dynamics and Chaos : With Applications in Physics, Biology, Chemistry, and Engineering, S.H. Strogatz, Perseus Press, 1994.
• Also a reading list of journal articles and conference proceedings will be distributed for this course.