Enhancing Breast Cancer Diagnosis: An Innovative Approach for Optimizing Neural Network Weights through Genetic Algorithms
Fatima Zahrae El-Hassani, Nour‐Eddine Joudar, Khalid Haddouch · 2024
Addressing the considerable health challenge of breast cancer, this research introduces an innovative strategy that combines genetic algorithms (GA) and back-propagation (BP) to enhance the effectiveness of breast cancer prediction models. From the UCI Machine Learning Repository which is widely recognized as the WDBC dataset the method conducts a global search with GA to identify an optimal set of connection weights, followed by a refined local search using BP. The results showcase superior performance, effectively addressing challenges like local minimum entrapment, and surpassing both the BP algorithm and a method relying solely on the genetic algorithm approach. This innovative technique marks a significant advancement in enhancing the accuracy of MLP-based breast cancer prediction models.