Fuzzy Logic

• Crisp logic: true/false absolut; contoh suhu >80°F = hot — tidak membedakan “sedikit di atas ambang” vs “jauh di atas”.
• Bahasa alami (“tall”, “very tall”) vagueness, bukan sekadar 0/1; fuzzy logic menoleransi data imprecise dan modeling non-linear.
• Fuzzy ≠ probability: probabilitas = likelihood kejadian; degree of membership = seberapa kuat objek masuk himpunan (bukan P(masuk set)).
• Contoh usia: Ann 28 → μ Young 0.8; Bob 35 → 0.3; Charlie 23 → 1.0 — overlapping membership functions Young/Middle/Old.
• Fuzzification: tentukan fuzzy set dan μ untuk input crisp (contoh tinggi 163 cm → vektor μ).
• Inference: evaluasi aturan fuzzy per rule; Composition: agregasi output semua rule; Defuzzification: nilai crisp final (mis. centroid).
• Keunggulan: mudah dipahami, berbasis pengalaman domain, tidak selalu butuh training seperti ML murni.

Alur sistem fuzzy ekspert tipikal dari slide terminology.

flowchart LR
  In["Input crisp"]
  Fz["Fuzzification"]
  Inf["Inference"]
  Comp["Composition"]
  Def["Defuzzification"]
  Out["Output crisp"]
  In --> Fz --> Inf --> Comp --> Def --> Out