Home
The software shows stimulation (pictures) and measures average changes to EEG. It then computes neural network machine learning model and uses it to choose stimulus which is more likely to cause a wanted target EEG state (only uses power spectrum values as the cheap customer EEGs have only few EEG channnels). This means dynamic stimulation of nervous system based on its temporal state where in psychiatry manipulations are always static and don't change/respond quickly to the current state of the nervous system. EEG-values are also processed using time-series classification algorithms to get meaningful states from the EEG.
This rather straightforward simple implementation manages to choose pictures that are likely to push a nervous system to some simple target state. However, in practice the methods are not sophisticated enough to steer/skillfully manipulate a nervous system state. The actual effects to CNS are then very small (18% change in EEG values).
Current version of software is here: https://github.com/cslr/neuromancer/releases/tag/v0.52a
Tomas Ukkonen
Here are examples of generated picture streams (adapted to my brain and nudges the brain toward same state as an EEG response to music) which should cause relaxedness (https://youtu.be/oPFCxLSF6XI and https://youtu.be/xSnEEur2t4o):
