From my previous projects and research work, I realized that recording EEG takes a lot a time, conditions maintenance and requires considerable investment, hence this project is aimed to reproduce brain waves corresponding specifically to some dominant constituent brainwave.
Features
- Generates EEG signals dominant over a specific frequency which corresponds to different brain states like Relaxed, Concentrated and Stress etc.
- Completely free from noise which is present during real EEG recordings such as noise due to a sensor, bandwidth and Motor activities of the subject
Workflow
- A small sample of previously recorded EEG data is fed which serves as the baseline for the EEG that is to be imitated.
- An optimization algorithm is run over a Neural Mass Excitation Model so that the potential value generated start imitating the input EEG.
- The model wights are altered by comparing the EEG features from the generated and required EEG. After the extent to which features match comes within a threshold range, it is safe to say that the EEG then generated imitates the input EEG.
Knowledge Gained
- Description features that help define and differentiates EEG Signals
- Implementing Optimisation algorithm over Biological Models to alter weights
- Implementing Reinforcement Learning for Optimisations.