4 ms·
You can also make your own EEG and train a classifier of your choice (heavily biasing the "do nothing" command) with sklearn or something. I've used off the sh
by krenoten 12y ago
You can also make your own EEG and train a classifier of your choice (heavily biasing the "do nothing" command) with sklearn or something. I've used off the shelf EEG's (emotiv epoc with the emokit library, there may be better options) to bind thought actions to moving a mouse, clicking, opening a web browser, but it was a bit noisy. I had to bind the act of clicking to the thought of a flavor, because it turned out I was thinking about "clicking" far before the mouse ever got to my intended destination. I didn't use it long enough to start thinking about the flavor before my mouse got there, but I'm curious if it would have turned into an arms race against my own mind eventually.
- Lambdanaut 12y agoI don't have a lot of experience(nil) in EEG, but it seems like the way you're doing it is the "unnatural" way to progress. We move our body parts as if with no thought at all. It's more like we "will" them to move rather than think the command "move!" and they move for us. They're connected to a different part of our brain, lower on the cortical hierarchy. Maybe you'd get better results if you instead pretended you physically were the computer, and the cursor was just an extremity that you could wave about and click with like one of your body parts. You'd bind the clicking and moving functions of the computer to this instead of verbal commands/thoughts. Or maybe I'm misunderstanding how this all works, and it's already done like I described. :)
- gcb0 12y agobut input is handled beautifully with the mechanical keyboard already. the neural interface was for output.