Big Data Imbalance Classification using Nonlinear Grasshopper Optimization Algorithm and Long Short-Term Memory with Rectified Linear Unit

Imbalanced data is common in the real world, and its classification remains one of the most researched topics.Existing techniques often suffer from imbalanced data, leading to overfitting and reduced classification performance.Therefore, a Nonlinear Decreasing Coefficient-based Grasshopper Optimization Algorithm (NGOA) and Long Short-Term Memory with Rectified Linear Unit (LSTM with ReLU) are proposed for big data imbalance classification.The traditional GOA, combined with a nonlinear decreasing coefficient, provides an adaptive optimization approach that improves search efficiency and solution quality in complex problems.Min-max normalization ensures that features are scaled properly, accounting for sensitivity to feature magnitude, thereby enhancing convergence and performance.The Synthetic Minority Oversampling Technique (SMOTE) is applied to generate synthetic data points, increasing the number of instances in minority classes to match those of the majority class through data synthesis.Metrics such as accuracy, precision, recall, F1-score, and Area Under the Curve (AUC) are used to evaluate the performance of the NGOA-LSTM with ReLU.The NGOA-LSTM with ReLU achieves accuracies of 96.00%, 99.80%, and 99.20% for the Pima, Yeast, and Heart datasets, respectively, compared to boosting approaches.