Research On The Concept Learning Based On DFS

Concept learning, one of the machine learning paradigms, has been widely used in data mining, pattern recognition, image processing and so on, and perfect results have been obtained in practice.Recently, there are many theories about concept learning such as Fuzzy Sets, Rough Sets, Boolean algebra, Classic Logic, and Granular Computing and so on, but none of them succeed to deal with dynamic fuzzy data.Theorefore, it is inevitable to use the theory of DFS into concept learning. The main achievements for 3 years are as follows:(1) Analyzed the feasibility of concept learning based on the theory of DFS, provided the model of representation about Dynamic fuzzy concept.(2) Proposed the model of concept learning based on dynamic fuzzy concept lattice, introduced the pre-processing of dynamic fuzzy formal background, the algorithms of constructing and reduction about it.(3) Proposed the model of concept learning based on dynamic fuzzy decision tree, introduced the construting and pruning about it, the algorithm of extraction about dynamic fuzzy concept rule.(4) We carry out some experiments about face recognition and classification in UCI repository to verify the effectiveness of our algorithms.