Enhancing EEG-Based Mental Task Classification: A Hybrid LSTM and CNN Approach
Zeyu Bai
University of California
Ruizhi Yang
University of California
Youzhi Liang
Massachusetts Institute of Technology
Abstract
This study presents an innovative approach to classifying mental tasks using electroencephalogram (EEG) signals. We developed a hybrid model combining Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks, leveraging the strengths of both to enhance classification accuracy. Our model employs a CNN decoder integrated with LSTM layers, demonstrating superior performance in validation and testing compared to traditional methods. With comprehensive experiments and analysis, we showcase the model's effectiveness in decoding EEG signals for mental task classification, offering significant advancements in brain-computer interface research. This paper details our approach, methodology, and the promising results obtained, underscoring the potential of our mixed LSTM-CNN model in neurocomputational applications.
