Deep Neural Network (DNN) for Outlier Detection in Heavy Metal Voltammetric Data

The majority of existing outlier detection methods are unable to adequately address the inherent complexity and variability of voltammetry data, particularly in the context of heavy metal detection. The effectiveness, precision, and adaptability of these methods are constrained, particularly when confronted with complex datasets. To address this gap, this study proposes a novel outlier detection method combining Isolation Forest, One-Class SVM, Local Outlier Factor (LOF), Z-Score, and Deep Neural Network (DNN) techniques. The proposed approach is designed to enhance accuracy by leveraging the unique strengths of each method in identifying outliers within complex data. Evaluation results demonstrate that the DNN method achieves the highest accuracy, reaching 99.87%, significantly outperforming other tested models. This robust performance suggests that the proposed method has substantial potential to improve the reliability of detecting heavy metal anomalies in voltammetric data. Specifically, the DNN method demonstrated an F1-score of 1.00 across multiple evaluation metrics, outperforming other models such as Isolation Forest (F1-score: 0.65) and LOF (F1-score: 0.75).