Harmful Brain Activity Classification Using Ensemble Deep Learning

Electroencephalography (EEG) is a crucial tool for monitoring electrical brain activity and diagnosing neurological conditions. Manual analysis of EEG signals is time-consuming and prone to variability, necessitating automated methods for accurate and efficient interpretation. This paper presents an ensemble learning approach for the automated classification of harmful brain activity from EEG data. The model combines the strengths of ResNet1d GRU, EfficientNetB0, and EfficientNetB1 to predict brain activity from spectrograms and raw EEG signals. By leveraging the unique capabilities of each model and employing a weighted averaging scheme, the ensemble achieves a competitive Kullback-Liebler divergence score of 0.3 on the testing dataset. This approach demonstrates the potential of ensemble learning to improve the accuracy and robustness of automated EEG signal classification, offering a valuable tool for early detection and diagnosis of neurological conditions.