فهرست نشریات علمی دسترسی آزاد ایران

This study introduces an enhanced Naïve Bayes classifier termed Adaptive Learning Naïve Bayes (ALNB) , designed to overcome the long-standing challenges of traditional models, including the conditional independence assumption and the zero-frequency problem. Building upon the Naïve Bayes Enrichment Method (NBEM, Koren et al., 2024[5]) , ALNB incorporates adaptive distribution selection using Kolmogorov–Smirnov and Anderson–Darling tests, and dynamic weighting to prioritize stronger classifiers. Experiments across medical, financial, and text datasets demonstrate that ALNB outperforms NBEM, Gaussian NB, Random Forest, and SVM in accuracy and robustness. This makes ALNB particularly effective in imbalanced and high-dimensional datasets. Potential applications include disease diagnosis, fraud detection, sentiment analysis, and spam filtering.