Paper Key : IRJ************273
Author: Amal Krishna N M,Dr. Priya S,Dr. Ashok Kumar T
Date Published: 07 Apr 2025
Abstract
Sampling in EEG-based Depression Detection System Using SMOTE is a project that addresses the challenge ofclass imbalance in depression detection datasets. This research explores the application of advanced samplingtechniques, including SMOTE, Gaussian SMOTE, and Adaptive Synthetic Sampling, to enhance the performanceof depression detection systems using EEG signals. The project leverages Graph Neural Networks (GNNs) forpredictive modeling, trained on EEG data that has been balanced using the aforementioned techniques. Byaddressing the issue of data imbalance, this study aims to improve classification accuracy and ensure robustperformance of the GNN model. Among the sampling methods investigated, Gaussian SMOTE demonstratedsuperior capability in generating realistic synthetic samples, leading to notable improvements in modelaccuracy. This research has the potential to advance the field of mental health diagnosis by enabling moreaccurate and reliable detection of depression, which can be instrumental in clinical settings and personalizedtreatment strategies. A concise conclusion highlights the findings and their implications for future research.Keywords: EEG-based Depression Detection, Sampling Techniques, SMOTE, Gaussian SMOTE, AdaptiveSynthetic Sampling, Graph Neural Networks (GNNs).
DOI Requested