George G. Lendaris

 
Professor (Systems Science and of Electrical & Computer Engineering)
Life Fellow, Institute of Electrical and Electronics Engineers
Fellow, International Neural Network Society
Ph.D. 1961, M.S. 1958, B.S. 1957 University of California, Berkeley

Research Interests

My research interests have long been focused in the arena now called Computational Intelligence, starting back in the 1960’s at the GM Research Labs in Santa Barbara, CA, following receipt of my Ph.D. My attention has been largely focused on the Neural Network (NN) aspect of Computational Intelligence, but has also included Fuzzy Logic.

Neural Networks is a massively-parallel computation methodology whose design is motivated by attributes of biological brain. My work has been mainly focused on methodology development. In the past 20-plus years, this has included (chronologically):

a)      Bridging symbolic AI with NN-based AI.

b)      Matching the structure/architecture of a NN to information-theoretic structural relations in data of a given problem context.

c)      Adaptive Critic methods for control system design.

d)     Experience based control and context discernment.

My current passion relates to the latter category, as I  believe to attain human-like capabilities in modern-day control systems, the equivalent of what is called ‘experience’ in humans will have to be implemented. Methodology development activities in this arena are based on what I am calling a “higher level” application of the method now known as Adaptive Dynamic Programming. One version is being called Higher Level Learning Algorithm (HLLA). The approach is described in the following two journal articles and Keynote Talk slides:

Lendaris, George G. (2008), "Higher Level Application of ADP: A Next Phase for the Control Field?" IEEE Transactions on Systems, Man, and Cybernetics--Part B: Cybernetics, Vol. 38, No. 4, p. 901-912.

Lendaris, George G. (2009), "Adaptive Dynamic Programming Approach to Experience-Based Systems Identification and Control," Neural Networks, Vol. 22(5), 822-832.

Lendaris, George G. (2011), "Higher-Level Application of Adaptive Dynamic Programming/Reinforcement Learning - a Next Phase for Controls and System Identification?" Proceedings from the 2011 Symposium on Adaptive Dynamic Programming & Reinforcement Learning, Paris, France, February. (Invited Keynote Talk)

Research in Neural Networks and Fuzzy Systems | Course Information