کنفرانس مهندسی برق ایران

صفحه اصلی / سی و سومین کنفرانس بین المللی مهندسی برق

CNN-LSTM model for Confusion Classification; using Single-Channel EEG

نویسندگان :

Amirhossein Aran¹ Zahra Ghanbari² Mohammad Hassan Moradi³

1- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران) 2- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران) 3- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران)

کلمات کلیدی :

CNN-LSTM model،Confusion detection،Single-Channel،Classification،Electroencephalography

چکیده :

Brain fog, characterized by decreased mental clarity and memory problems, is a common phenomenon that affects both healthy individuals and those with mental disorders. Detecting and alleviating brain fog is essential for activities such as driving and online learning. Electroencephalography (EEG) is an effective method for monitoring brain activity and identifying confusion. “Confused student EEG brainwave data” from the Kaggle challenge “EEG Brainwaves for Confusion,” consists of single-channel EEG recordings of ten students exposed to easy and difficult video stimulation is used in this paper. We take benefit from a powerful interpretable hybrid convolutional neural network and long short-term memory (CNN-LSTM) model. CNN-LSTM processes EEG signals through CNN for feature extraction, followed by the LSTM for temporal analysis and data classification using a Softmax layer. This is the first time that CNN-LSTM is used for single channel EEG confusion classification. The model achieves 93.44% accuracy, 92.53% precision, 94.79% recall, and 93.65% F1 score. These results are dramatically higher than previous studies, which demonstrates the potential of CNN-LSTM model in effectively distinguishing confusion and non-confusion states in EEG signals, providing a promising method for real-time confusion detection.