New S-norm and T-norm Operators for ALM

Please enable javascript in your browser.

Kiaei Khoshroudbari, Ali Akbar | 2088

555 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 39013 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Bagheri Shouraki, Saeed
Abstract:
Active Learning Method (ALM) is a soft computing methodology which has been used for modeling and control. Unlike the good results of ALM’s ability, there is not any analytic expression to prove its power, because ALM’s main operators don’t satisfy fuzzy S-norm & T-norm properties. This problem considered from three viewpoints: Mathematics, Geometry and fuzzy sets. In this paper, we try to introduce two new operators based on morphology to satisfy three conditions: at first, they must be fuzzy S-norm and T-norm to confirm the committal of fuzzy set theory. Secondly, they must satisfy Demorgans law, because they are complement of each other. And finally, they resemble the main operators of ALM, because support ALM’s operation well. We use them instead of ALM’s original operators and prove the Fuzzy S-norm and T-norm for these two new operators. Moreover we consider new operators from three viewpoints said above
Keywords:
Reinforcement Learning ; Ink Drop Spread (IDS)Operator ; Thinning ; S-Norm Operator ; T-norm Operator ; Center of Gravity (COG) ; Thickening