Title :
Generalized entropy and minimum system complexity
Author :
Pandelidis, Ioannis
Author_Institution :
Dept. of Mech. Eng., Maryland Univ., College Park, MD, USA
Abstract :
The problem of characterizing information processing systems in terms of their entropic complexity is examined. Automatic control systems and the design/manufacturing process are seen as such systems, so that the results of such research are intended primarily for this area. The concept of generalized entropy is introduced in order to unify the treatment of continuous and discrete signals and the corresponding information processing systems. Given a level of entropic complexity of the reference and disturbance inputs to a system, the author derives minimum entropic complexity requirements for that system. It is assumed that the entropic system complexity that is defined by the author is monotonically related to cost. Based on this assumption and a previously proposed design/manufacturing axiom, the system having minimum entropic complexity is derived
Keywords :
control system analysis; information theory; manufacturing processes; design/manufacturing process; entropic complexity; entropy; information processing systems; minimum system complexity; Adaptive systems; Artificial intelligence; Artificial neural networks; Associative memory; Cost accounting; Entropy; Fuzzy logic; Information processing; Learning systems; Neural networks;
Journal_Title :
Systems, Man and Cybernetics, IEEE Transactions on
