1DCNN Mental Workload Classifier

A machine learning project that classifies cognitive workload levels from EEG signals using Convolutional Neural Networks (CNN).

📋 Overview

This project analyzes EEG brain signals to determine whether a person is experiencing high or low cognitive workload. Using the STEW Dataset, we built a CNN model that can classify workload levels with visualization of brain activity patterns.

🎯 Project Goals

• Binary Classification: Distinguish between low/medium workload (ratings 4-6) and high workload (ratings 7-9)
• EEG Visualization: Generate signal plots and heatmaps to understand brain activity patterns
• Model Performance: Achieve reliable classification accuracy using deep learning

📊 Dataset

STEW Dataset (Sustained Attention to Response Task EEG Workload Dataset)

• 14 EEG channels (Emotiv headset layout)
• 64 time points per sample
• Cognitive workload ratings from 4-9
• Binary classification target (0: Low/Medium, 1: High)

🧠 Model Architecture

CNN Model Features:

• 3 Convolutional layers with MaxPooling
• Batch Normalization and Dropout for regularization
• Dense layers for final classification
• Sigmoid activation for binary output

📁 Repository Structure

├── Mental_workload_model.ipynb    # Main Jupyter notebook with complete code
├── Project_Report.pdf              # Detailed project report
├── PRESENTATION_1DCNN-Mental-Mental-Workload-Classifier.pdf              # Project presentation slides
└── README.md                       # This file

🚀 Quick Start

1. Clone the repository

   git clone https://github.com/your-username/1DCNN-Mental-Workload-Classifier.git
   cd 1DCNN-Mental-Workload-Classifier

2. Open the notebook

   • Upload to Kaggle/Google Colab or run locally
   • Download the STEW Dataset files: dataset.mat, rating.mat, class_012.mat

3. Run the notebook

   • Follow the cells step by step
   • The notebook includes data preprocessing, model training, and visualization

📈 Key Features

• Data Preprocessing: Z-score normalization and proper channel alignment
• CNN Training: Optimized for EEG temporal patterns
• Visualizations:
  • EEG signal curves for individual channels
  • Heatmaps showing signal intensity across time and channels
• Performance Metrics: Accuracy, Precision, Recall, F1-score, ROC-AUC

🛠️ Technologies Used

• Python: Core programming language
• TensorFlow/Keras: Deep learning framework
• NumPy & Pandas: Data manipulation
• Matplotlib: Visualization
• scikit-learn: Model evaluation
• SciPy: Signal processing

📊 Results

The model successfully classifies cognitive workload levels with detailed performance metrics included in the notebook. Visualizations help understand which brain regions are most active during high cognitive load.

📚 Documentation

• Complete Code: eeg-cognitive-workload.ipynb
• Detailed Analysis: project-report.pdf
• Presentation: presentation.pptx

🔮 Future Improvements

• Multi-class classification for more granular workload levels
• Feature importance analysis using explainability tools
• Real-time classification capabilities
• Expanded dataset for better generalization

📄 License

MIT License - Feel free to use and modify this project.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request or open an Issue for any suggestions or improvements.

---

Note: Make sure to download the STEW Dataset files before running the notebook. The dataset is publicly available for research purposes.