@article{M731D4905,
title = "Analysis on Underwater Channel by Using Shapley Additive Explanations",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2025",
issn = "1226-4717",
doi = "10.7840/kics.2025.50.7.1007",
author = "Jongseok Kim, Ho-Shin Cho, Ohyun Jo",
keywords = "Underwater Communication, XAI, SHapley Additive exPlantaions, Feature Selection",
abstract = "This study explores the limitations of relying solely on Signal-to-Noise Ratio (SNR) for Bit Error Rate (BER) prediction in underwater communication environments and underscores the critical role of eXplainable Artificial Intelligence (XAI). By employing SHapley Additive exPlanations (SHAP), the relationship between SNR and BER is thoroughly analyzed, highlighting the inadequacies of SNR as the sole predictive feature. To address these challenges, SHAP-based feature selection is utilized to identify key factors, which are subsequently employed to train machine learning models. The results demonstrate a marked improvement in prediction accuracy over traditional methods, affirming that the integration of SHAP-driven feature selection significantly enhances model performance."
}