Diagnosing Breast Cancer Based on the Adaptive Neuro-Fuzzy Inference System.

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• Accurate data classification is crucial in machine learning.
• Existing methods like RBFNN and ANFIS have limitations in certain datasets.
• Hybrid approaches offer potential for enhanced classification performance.

Purpose of the Study:

• To introduce a novel hybrid neuro-fuzzy classifier (HNFC) for improved input data classification.
• To evaluate the performance of the proposed HNFC against established supervised classification techniques.
• To demonstrate the effectiveness of HNFC on benchmark datasets, including the breast cancer dataset.

Main Methods:

• Input data fuzzification using a generalized membership function.
• Feature selection via statistical correlation after missing data imputation and normalization.
• Integration of fuzzy logic and neural network for classification and performance evaluation.

Main Results:

• The proposed HNFC technique demonstrated high classification accuracy across ten benchmark datasets.
• Specifically, the HNFC achieved 86.2% classification accuracy on the breast cancer dataset.
• Performance was evaluated using accuracy and error rate metrics, showing superiority over RBFNN and ANFIS.

Conclusions:

• The novel hybrid neuro-fuzzy classifier (HNFC) offers a significant advancement in data classification accuracy.
• HNFC effectively handles data preprocessing steps including missing value imputation, feature selection, and normalization.
• The method shows promising results, particularly for complex datasets like the breast cancer dataset, outperforming existing approaches.