European Proceedings Logo

Self-Configuring Evolutionary Algorithms Based Design of Hybrid Interpretable Machine Learning Models

Table 2: Databases of fuzzy rules describing the decision-making process of the ANN

Classification task Rules
BTS Rule 1: «if (Recency (months) is S) and (Frequency (times) is L) then Consent»
Rule 2: «if (Frequency (times) is S) and (Monetary (c.c. blood) is S) and (Time (months) is L) then Refusal»
BCW Rule 1: «if (mean compactness is L) and (worst texture is M) and (worst perimeter is M) then Benign»
Rule 2: «if (mean concave points is S) and (worst perimeter is S) then Malignant»
CR Rule 1: «if (loan is S) then Refusal»
Rule 2: «if (age is L) then Refusal»
Rule 3: «if (income is S) and (age is M) and (loan is L) then Consent»
BA Rule 1: «if (variance of Wavelet Transformed image is S) and (curtosis of Wavelet Transformed image is S) then Genuine»
Rule 2: «if (variance of Wavelet Transformed image is L) then Fake»
Rule 3: «if (skewness of Wavelet Transformed image is L) then Fake»
UKM Rule 1: «if (The exam performance of user for related objects with goal object is M) and (The exam performance of user for goal objects is M) then Low»
Rule 2: «if (The exam performance of user for goal objects is M) then Middle»
Rule 3: «if (The exam performance of user for related objects with goal object is M) and (The exam performance of user for goal objects is L) then High»
Rule 4: «if (The exam performance of user for related objects with goal object is S) and (The exam performance of user for goal objects is S) then Very Low»
< Back to article