Examples

This section provides four end-to-end examples covering the main workflows supported by Neuro-Fuzzy Toolbox. Each example includes data loading and preprocessing, model instantiation, parameter initialization, training, and evaluation. The examples are ordered roughly by complexity, from a standard classifier built with a built-in training algorithm to structural adaptation with SONFIS on regression data.

The datasets used are publicly available through the UCI Machine Learning Repository or generated synthetically. Data preprocessing, train/validation/test splitting, and DataLoader construction follow standard scikit-learn and PyTorch conventions.

• Example 1 covers multiclass classification on the Iris dataset using an h_ANFIS model trained with the Basic Optimizer Training Algorithm, and demonstrates post-training rule inspection with RulesAnalyzer.

• Example 2 covers multiclass classification on the Glass Identification dataset using a rule_reduced_ANFIS model, combining an initial gradient-based training phase with a custom greedy rule-growing procedure built on the low-level API.

• Example 3 covers binary classification on the Heart Disease dataset using a rule_reduced_ANFIS model trained with SONFIS for structural adaptation.

• Example 4 covers regression of a noisy 3D surface using a rule_reduced_ANFIS model trained with SONFIS, including surface visualizations of the target function and model predictions.

• Example 1: Multiclass Classification on Iris Dataset
  • Imports and reproducibility
  • Data
  • DataLoaders
  • Model
  • Learning algorithm
  • Evaluation
  • Rule structure analysis
• Example 2: Multiclass Classification on Glass Identification Dataset
  • Imports and reproducibility
  • Data
  • DataLoaders
  • Model
  • Initial training
  • Initial evaluation
  • Custom strategy: greedy rule-growing
  • Final evaluation
• Example 3: Binary Classification on Heart Disease Dataset using SONFIS
  • Imports and reproducibility
  • Data
  • DataLoaders
  • Model
  • Learning algorithm
  • SONFIS
  • Evaluation
• Example 4: Regression on a 3D Surface using SONFIS
  • Imports and reproducibility
  • Visualization helper
  • Data
  • DataLoaders
  • Model
  • Learning algorithm
  • SONFIS
  • Evaluation