Akademik Veri Yönetim Sistemi
4th International Conference on Intelligent and Fuzzy Systems (INFUS), Bornova, Türkiye, 19 - 21 Temmuz 2022, cilt.504, ss.73-81
Discrimination of two linearly inseparable sets using hyperplane(s) is considered one of the most successful classification methods. There is an expanding body of literature that realize the significance of classification of data accurately. Particularly, medical applications of classification provide meaningful early diagnosis results. However, medical datasets in real-world applications possess some noise and uncertainty. Therefore, dealing with uncertainty is a critical factor for accurate diagnosis. To overcome mentioned drawbacks, this study presents a hybrid fuzzy rule-based robust linear programming (RLP) and h-polyhedral separation (h-PolSep) approaches for breast cancer diagnosis through several consequent stages. In the proposed two-stage model, firstly the data is relabeled according to the fuzzy rule-based system, then the new outputs and original input values are classified using RLP and h-PolSep methods. Input fuzzification, generating membership functions, extracting fuzzy rules, and output defuzzification are examined in detail. A frequently used real-world medical dataset from the UCI Repository: the Wisconsin breast cancer is employed to evaluate the effectiveness of the classification using a number of metrics. Finally, the results demonstrate that the proposed approach effectively handles medical data classification problems.