0
$\begingroup$

I am working on a project that is basically a game for motor-paralyzed people.

It should take an EEG signal from FP1 channel of brain and then after processing it should generate command for the game (like left,right, stop, fire, jump etc).

In the start stage of the project, I want to train my ANN on data that has already been gathered.

I am new to the signal processing field. The dataset I found over internet is unlabeled data (i.e. it is just a signal data; it does not tell about output of the specific data row).

So can somebody, who has working experience with EEG signal, tell me what I should do? More specifically, I am unable to understand the signal.

$\endgroup$
  • $\begingroup$ well, if the data came unlabeled, it's no use for you, as you'd have zero basis to validate it, even if you could understand it. And since you're not even linking to the data set, we can't even understand it. Hence, what you need to do: Find a better data set, or learn how to record and preprocess EEG signals (there's plenty of literature!) and record your own data set. $\endgroup$ – Marcus Müller Nov 12 at 19:30
  • $\begingroup$ @MarcusMüller If I record My own dataset it is still unlabeled data(as it is a signal readings) so how can I classify if this is unlabeled data? Moreover what I have seen is, the EEG data is just a signal. It is not like the other conventional datesets which are being used for classifiers. So it is making my life difficult. I need help in this specific part. $\endgroup$ – majid bhatti Nov 12 at 21:42
  • 1
    $\begingroup$ why would you not label your data while recording it? There's no "helping you in this specific part": There's nothing specific about your problem. The ML classifier doesn't care whether your data is a signal or a set of stock prices or just random gibberish. If you don't have labels nor have knowledge on the distribution, you don't have anything to train with. There's no information in your system, and you can't "invent" info. $\endgroup$ – Marcus Müller Nov 12 at 22:15
  • $\begingroup$ If it's unlabeled how does the classifier know you're thinking about going up, not going to the toilet? Some AI algorithms may be able to identify different clusters in the data, but you won't know what they actually are. Imagine that in another context: you'd be making a controller where the buttons are randomized and not labeled. "This one (on the right) means go left. The one on the left is actually the main circuit breaker for the building, don't touch it. To go right you have to turn the microwave on. To use the microwave, this lightswitch is the 4 key and we haven't found the others." $\endgroup$ – user253751 Nov 13 at 12:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.