Skip to content

Latest commit

 

History

History
5 lines (3 loc) · 1.37 KB

README.md

File metadata and controls

5 lines (3 loc) · 1.37 KB

fcmappers

Serial implementation of or hybrid classifier proposed in G.A. Papakostas and D.E. Koulouriotis. Classifying Patterns Using Fuzzy Cognitive Maps. In: M. Glykas (Ed.): Fuzzy Cognitive Maps, STUDFUZZ 247, pp.291–306, 2010. Springer-Verlag Berlin Heidelberg. In this case, the FCMper2 operates at a second stage. In the first stage a black box of any type and structure is trained with the training data set, for a specific application. Afterwards, the trained black box by working in generalization mode, tries to recognize the training patterns. The decisions of the black box constitute the values of the input concepts of the FCMper2.

Parallel implentation of hybrid classifier proposed in G.A. Papakostas and D.E. Koulouriotis. Classifying Patterns Using Fuzzy Cognitive Maps. In: M. Glykas (Ed.): Fuzzy Cognitive Maps, STUDFUZZ 247, pp.291–306, 2010. Springer-Verlag Berlin Heidelberg. In FCMper3 the input nodes are not only the blackbox decisions, but the input feature vector values as well. In this case, the proposed model attempts to model a supervisor classifier at a second stage, which takes the decisions of the first stage classifier (black box) for specific input features and maps them to more accurate ones. This structure combines the advantages of the FCMpers 1 and 2, by managing the whole available information coming from both the black box and the input data.