Takagi-Sugeno-Kanga Fuzzy Fusion In Dynamic Multi-Classifier System
Maciej Krysmann · Proceedings of the World Congress on Electrical Engineering and Computer Systems and Science · 2016
In this paper, the approach to implementation of Takagi-Sugeno-Kanga fuzzy system into the Dynamic Ensemble Selection multi-classifier.Paper presents DES system with its working idea, provides in-depth information about that system.Dynamic creation of classifier ensemble, which selects classifiers for particular classified object x has proven its advantages.It is shown that described TSK system can improve classification quality even better, even in situation in which base classifiers are not fully trained.Proposed rule set and for TSK system is described.Paper presents complete algorithm with pointing all phases of work.Experimental study presents positive results and prove proposed system advantages basing on well known UCI Machine Learning benchmark databases.Paper also is discussing real life situation in which system can be used, however also points out classification time increase.