4.3.4 A Change in the Measurements -- 4.3.5 One-Pass Design: Non-singleton Fuzzification -- 4.3.6 Derivative-Based (BP) Design: Non-singleton Fuzzification -- 4.3.7 Final Remark -- 4.4 Case Study: Knowledge Mining Using Surveys -- 4.4.1 Methodology for Knowledge Mining -- 4.4.2 Survey Results -- 4.4.3 Determining Type-1 Fuzzy Sets from Survey Results -- 4.4.4 What Does One Do with a Histogram of Responses? -- 4.4.5 Averaging the Responses: Consensus FLAs -- 4.4.6 Preserving All of the Responses -- 4.4.7 On Multiple Indicators -- 4.4.8 How to Use an FLA -- 4.4.9 Connections to the Perceptual Computer -- 4.5 Case Study: Rule-Based Classification of Video Traffic -- 4.5.1 Compressed Video Traffic -- 4.5.2 High-Level Video Classification Problem -- 4.5.3 Selected Features -- 4.5.4 MFs for the Features -- 4.5.5 Rules and Their Parameters -- 4.5.6 Computational Formulas for the RBC -- 4.5.7 Optimization of Rule Design Parameters -- 4.5.8 Testing the FL RBC -- 4.5.9 Results and Conclusions -- 4.6 Case Study: Fuzzy Logic Control -- 4.6.1 Early History of Fuzzy Control -- 4.6.2 What Is a Type-1 Fuzzy Logic Controller (FLC)? -- 4.6.3 Fuzzy PID Control -- 4.6.3.1 Background -- 4.6.3.2 General Structure of Fuzzy PID Controller -- 4.6.3.3 Conventional and Fuzzy PID Controller Design Methods -- 4.6.3.4 Simulation Results (T1-FPID Versus PID) -- 4.7 Case Study: Explainable Type-1 Fuzzy System -- 4.7.1 Computations Common to Both Fuzzy Systems -- 4.7.1.1 Firing Levels for the Active Rules -- 4.7.1.2 Similarities -- 4.7.2 Mamdani with Centroid Defuzzification -- 4.7.2.1 Computation of yc (2.4, 5.4, 9) -- 4.7.2.2 Explaining yc (2.4,5.4,9) -- 4.7.2.3 Quality of Explanation -- 4.7.3 Mamdani with COS Defuzzification -- 4.7.3.1 Computation of yCOS(2.4,5.4,9) -- 4.7.3.2 Explaining yCOS(2.4,5.4,9) -- 4.7.3.3 Observations -- 1.1 Count of MF Parameters. |