A Comprehensive Study of Classical Machine Learning and Deep Learning in Subject-Independent BCI Using Adaptive SNR-Weighted DWT Artifact Elimination

1. Proposed SA-DWT: SNR-based adaptive EEG denoising for each subject and channel.
2. Preserves useful brain signals while removing noise more effectively.
3. Improves subject-independent, zero-calibration MI-BCI performance.
4. Validated with LOSO-CV across five ML/DL classifiers.