Optimizing the Backpropagation Neural Network with the Whale Optimization Algorithm

Amidst the rapid evolution of large data technology, the need to swiftly and accurately filter information from large datasets has become paramount. This paper introduces a novel machine learning model denominated as 'WOA-BP', which synergistically aligns the Whale Optimization Algorithm (WOA) with the Backpropagation (BP) Neural Network. The newly proposed model leverages the WOA's rapid convergence and robust global search capabilities to optimize the weight and biases of the BP neural network. To assess its performance, the WOA-BP model is benchmarked against five industrial models using publicly available datasets. Notably, the WOA-BP algorithm achieves an outstanding regression prediction score (R2 value) of 0.954 along with an impressive classification accuracy of 95.56%, both of which are the highest among all models evaluated. It is worth highlighting that the WOA-optimized models (e.g., WOA-Decision Trees and WOA-Bayesian) consistently outperform their counterparts (e.g., the conventional Decision Trees and Bayesian models). These findings highlight the promise of the WOA for future applications in enhancing machine learning models and addressing complex optimization challenges.