Deep Convolutional Neural Network Regularization for Alcoholism Detection Using EEG Signals.

Area of Science:

• Neuroscience
• Machine Learning
• Biomedical Engineering

Background:

• Alcoholism disrupts the brain's neuronal system, causing detectable malfunctions.
• Electroencephalogram (EEG) signals can capture these neural disturbances.
• Classifying alcoholism using minimal EEG data presents a significant challenge due to signal complexity.

Purpose of the Study:

• To accurately classify alcoholism using electroencephalogram (EEG) data from a minimal number of electrodes.
• To investigate the effectiveness of Convolutional Neural Networks (CNNs) for EEG-based alcoholism detection.
• To optimize a CNN model for improved accuracy and performance metrics.

Main Methods:

• Applied a baseline Convolutional Neural Network (CNN) model to raw EEG data.
• Optimized the CNN model using dropout, batch normalization, and kernel regularization techniques.
• Evaluated model performance on the University of California at Irvine Machine Learning (UCI-ML) EEG dataset.

Main Results:

• Achieved an average accuracy of 98% in classifying alcoholism from EEG data.
• Demonstrated superior performance compared to a baseline CNN model and other approaches.
• Showcased the effectiveness of optimization techniques in enhancing classification accuracy.

Conclusions:

• Optimized CNN models can accurately detect alcoholism from EEG signals with high precision.
• Techniques such as dropout and batch normalization are crucial for improving CNN performance in this application.
• The developed approach offers a promising tool for diagnosing alcohol use disorder non-invasively.