Github

Data Source

We used the SEED Dataset V, provided by the Department of Computer Science at Shanghai Jiao Tong University. The dataset was collected during experiments where participants were shown video clips designed to evoke emotions such as happiness, fear, sadness, disgust, and boredom. These EEG recordings served as the foundation for our analysis and modeling.

Tools and Technologies

• Python: For data analysis, model implementation, and visualization.

• Jupyter Notebook: Used for data exploration, analysis, and iterative development.

• Pandas & Seaborn: For data manipulation and creating insightful visualizations such as heatmaps and trend plots.

• Recurrent Neural Network (RNN): Built to detect temporal patterns in EEG data.

• Docker: To containerize the application, ensuring consistent and reproducible deployment.

• Streamlit: To create an interactive and user-friendly web application for visualizing and testing the model.

Methodology

1. Data Preprocessing:

   • Processed raw EEG signals, removing noise and extracting relevant features.

   • Utilized Pandas for data cleaning, transformation, and exploration.

2. Model Development:

   • Developed a Recurrent Neural Network (RNN) for analyzing time-series EEG data.

   • Used Support Vector Machines (SVM) as a baseline for comparison.

3. Data Visualization:

   • Employed Seaborn to create visually appealing graphs and plots to analyze data trends and evaluate model performance.

4. Deployment:

   • Containerized the app with Docker for portability and ease of setup.

   • Built an interactive dashboard using Streamlit, allowing users to visualize the EEG data, explore emotional states, and test the model’s predictions.

Results

• The RNN model achieved 68% accuracy in predicting participants’ emotional states based on EEG brainwave patterns.

• The Streamlit app provides a dynamic interface for exploring the data and results, making the project accessible to both technical and non-technical audiences.

Key Takeaways

This project highlights our ability to:

• Analyze and preprocess large datasets using tools like Pandas.

• Develop and fine-tune machine learning models for real-world applications.

• Create compelling visualizations using Seaborn.

• Deploy scalable applications with Docker and build interactive dashboards with Streamlit.

Practical Applications

This project underscores the potential of combining neuroscience, machine learning, and web applications to create tools for mental health monitoring, emotion-aware technologies, and brain-computer interfaces.