Neuroscience / Scientific Machine Learning
EEG Vector Representation System
A vector representation pipeline for comparing Parkinson’s and healthy-patient EEG signals.
- Institution
- Independent Project
- Role
- Research Developer
- Timeline
- 2025
- Status
- Complete
Wilson–Cowen signals transformed with CEBRA and stored as searchable vector embeddings.
PythonCEBRAChromaDBEEG
01EEG signals02CEBRA encoder03Vector embeddings04Comparative analysis
Neurological signals are high-dimensional and difficult to compare directly.
Generate excitatory-inhibitory patterns with the Wilson–Cowen model, encode EEG data with CEBRA, and store the representations in a vector database.
Dhruva built the simulation, embedding, storage, visualization, and comparison workflow.
An offset-10 model was trained to identify differences between Parkinson’s and healthy-patient EEG data.
Representation design determines which neurological differences become legible to downstream systems.