One way to decrease noise is to average across trials , however as a result less trials will be available for decoding . I wonder whether it will be OK to average across trials with replacement, when applying linear classifier , of course for training data and testing data separately .
Thank you very much
All trials are for the same event you are trying to decode, so why do you want to use some of them?
if you average your trials and train a classifier with avraged trails, your classifier could not work well in classifing new trails.
Imagine you have a new single trial and want to classify it.
Here is the question: How can you average one trail?!
so you have to give a trail that isn't averaged!
then your classifier won't classify correctly.
But in p300 it's sth normal to average some trails. But not the way you think.
Maybe this research gives insight into the subject.
https://www.sciencedirect.com/science/article/abs/pii/S016787609600743X
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training