I record the emg signal of 6 motion from 10 subjects. 2 channel had been used in data acquisition and each motion is repeated for 10 times. For classification, is that mean i need to use all 10 trials of each motion from every subject? In short i using total 1200 (6motion x 10repetition x 2channel x 10subjects) emg data for classification?
if you are going to classify 6 motion, you have to process data for each subject and you should divided the acquire data into two sets one for training and one for testing or may be you can use the cross validation . Also you should know how many features you will use to build your feature vector then you may need use feature reduction . then by the end of the day, you should calculate the average classification accuracy of the ten subjects. I think you should read more about EMG based pattern recognition in the literature. I've already attached some useful links for you before, but you have to find more.
Dear Maged Al-Quraishi,
But it is possible to use the data of ten subject at once in classification. Since each subject might produce different emg signal and if i want to apply in real system i need to use all data of ten subject but not average classification?
A real system using EMG would be specific to an individual, and would need to be trained for that individual using that specifc individual's EMG readings. This is similar to how you would not use an "average EEG" signal to develop a single brain-computer interface for multiple subjects. Each BCI needs to be trained on the signals of one specific subject and will only have decent classification accuracy for that individual
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing