Fuzzy Rule Based Knowledge Discovery And Representation

In this era of digital revolution, Knowledge Discovery from data is becoming increasingly important. Knowledge representation and reasoning has long been recognized as a central issue in Knowledge Discovery. Fuzzy rule-based system as a well-known Knowledge Discovery technique can effectively deal with knowledge representation and reasoning (fuzzy rule denoting knowledge and fuzzy logic reasoning knowledge). Fuzzy rule-based system can effectively solve various tasks of Knowledge Discovery including regression, clustering, classification, prediction, etc. Fuzzy rule-based system has many advantages including high classification accuracy and easily understood by human being. This thesis focuses on fuzzy rule-based system to resolve some problem about Knowledge Discovery including fuzzy decision tree, fuzzy rule based clustering, and fuzzy rule based data granulation. Main topics include:1. This thesis presents a new architecture of a fuzzy decision tree based on fuzzy rules-Fuzzy Rule based Decision Tree (FRDT). An association rules extraction algorithm (AREA) is proposed which is guided by a criterion of Fuzzy Confidence. The building of FRDT is realized the proposed association rules extraction algorithm AREA. In contrast with traditional decision trees in which only a single feature (variable) is taken into account at each node, the node of the proposed decision trees involves a fuzzy rule which involves multiple features. Fuzzy rules are employed to produce leaves of high purity and minimise the size of the trees. FRDT can effectively overcome the lack of semantic interpretation of the oblique decision trees which utilize hyperplanes as decision functions. We have applied FRDT on UCI machine learning datasets and analyzed the performance of FRDT. The comparison is carried out with regard to'"traditional" decision trees such as C4.5, LADtree, BFTree, SimpleCart, and NBTree. The results of statistical tests have shown that the proposed FRDT exhibits the best performance in terms of both accuracy and the size of the produced trees.2. In the framework of Axiomatic Fuzzy Set (AFS) theory, this thesis proposes a new approach to data clustering-AFS fuzzy clustering. Firstly, some simple concepts are selected to describe every sample (instance, pattern). Then these simple concepts are aggregated by AFS theory to form complex concepts serving as a description of this sample. Finally, the samples with the same or similar descriptions are regarded as forming a single cluster. The nature of each cluster is described by the aggregations of the descriptions of the typical samples in this cluster, and in this sense the cluster comes with a clearly articulated semantics. The effectiveness of the proposed approach is demonstrated by UCI machine learning datasets, and the obtained results show that the performance of the clustering is comparable with other fuzzy rule-based clustering methods, and benchmark fuzzy clustering methods FCM and K-means.3. A rapid data granulation method (FRCGC) is proposed to give suitable descriptions for all information granules. Firstly, some features are selected by a proposed unsupervised feature selection method. After that, calculate each sample's description by fuzzy rules on the remaining features. Finally, exemplar descriptions are selected from sample's descriptions according to their importance, and data granulation (the clustering procedure) is guided by the selected exemplar fuzzy descriptions. The definition of sample's description can be changed according to the domain knowledge, so that the proposed method can be easily improved to deal with complex actual problems. The experimental results show that our proposed model is able to discover fuzzy IF-THEN rules to obtain the potential granules effectively.