- Go to "Data Generation" page
- Set duration (10 seconds is good for testing)
- Choose brain states like "relaxed", "focused", "excited"
- Click "Generate Brainwave Data"
- You'll see realistic EEG signals with different patterns for each brain state
- Go to "Signal Processing" page
- Apply filters to clean the brainwave data:
- Bandpass filter (1-50 Hz) removes unwanted frequencies
- Notch filter removes electrical interference
- Artifact removal cleans up noise
- Click "Process Signal" to see before/after comparison
- Go to "Tokenization" page
- Choose how to convert brainwaves to tokens:
- Amplitude bins: Groups similar signal strengths
- Frequency bands: Groups by brain wave types (alpha, beta, etc.)
- Pattern-based: Finds repeating patterns
- Clustering: Groups similar brain states
- Click "Create Tokenizer" then "Tokenize Data"
- See how your brainwaves become numbered tokens (like 1, 5, 12, 7...)
- Go to "Model Training" page
- The AI learns patterns from your brainwave tokens
- Choose model settings (LSTM works well)
- Click "Train Model" - watch the loss decrease as it learns
- Go to "Inference" page
- Feed new brainwave data to the trained model
- The AI generates responses based on the brainwave patterns it learned
- Different brain states produce different types of responses
- Go to "Analytics" page
- See detailed analysis of your brainwave patterns
- Understand what the AI learned from your brain data
Instead of training on words, this AI trains on brainwave patterns. Each unique brainwave pattern becomes a "token" that the AI can learn from and use to generate responses.