Empowering white shark optimizer for dimensionality reduction with case study of apple disease prediction

Feature selection (FS) plays a crucial role in developing classification models by reducing the number of features used while improving their predictive power. It is a challenging problem that can be viewed as an NP-hard optimization task. To tackle this problem, powerful wrapper-based metaheuristic algorithms are employed, as they have the ability to search for nearly optimal feature subsets in the vast search space. However, these algorithms often face challenges such as getting trapped in local optima and striking a balance between exploration and exploitation. To address these challenges, this study proposes an improved version of the white shark optimizer (WSO) called the improved WSO (IWSO). The IWSO incorporates two efficient strategies, namely opposition-based learning (OBL) and Gaussian mutation (GM), to overcome the limitations of the original method. OBL enhances exploration by considering opposite solutions, while GM prevents premature convergence and improves the exploitation capabilities of the algorithm. The effectiveness of the proposed IWSO is evaluated using various benchmark datasets and assessed using standard evaluation metrics. The results of the experiments demonstrate that the IWSO is capable of discovering new optimal solutions across different test cases. Furthermore, the proposed algorithm is applied to a real-world problem involving the identification of apple diseases, further validating its effectiveness.