I’m a computer science student and fascinated about neuroscience. I would like my bachelors final project to revolve around the topic. I would like to apply machine learning / data analysis on eeg raw data to read information from my brain.
Specifically, what I would like to do is think of concepts or words and detect that with my code. For example, I could think of an apple or a car, and the program would be able to detect what my thoughts were (obviously within a restricted domain).
I saw examples of eeg gadgets that detect simple concepts like up, down left and right, so I’m guessing it is at least somewhat feasible. What I wonder is up to which extent? What do you think?
Maybe the question is a bit stupid, but I would still appreciate an answer since I am a complete beginner in this field and I need to know if the investment is worth it or not.
This is interesting.
Do you think it will work to detect whole words at once, instead of letter by letter? Say, from a pool of around 100 words? I guess the answer depends on the amount of precision the EEG provides.
So, my next important question would be: which EEG should I get to work on this? Are some of the popular options like Emotiv or Bitbrain, or should I contact Cognionics instead?
Some type of speller is the way to go. Here are some other links. Cognionics is very expensive. g.tec has a nice speller, but it is also way expensive. OpenBCI has good experience in this area,