[Eeglablist] The data pruned by ICA

[Arnaud Delorme]

Dear Fan,

the values in EEG.data should be strictly identical to EEG.icawinv*EEG.icaact (assuming you used all the data channels when you performed ICA) or EEG.icawinv*EEG.icaact(:,:) for data epochs. The difference might be 10^-17 because of numerical limitations of Matlab but probably not more.

Best,

Arno

On Apr 24, 2011, at 5:36 AM, fanz at psy.gla.ac.uk wrote:

> Hi,
> 
> I run ICA and get a new data set which only contains the components  
> represent motor imagination, and I want to use this new dataset in the  
> next step of classifciation. My question is, the data I get by   
> icawinv*icaact is slightly different from EEG.data which has been  
> pruned by ICA. So could someone tell me what's wrong?
> 
> Best wishes
> 
> Fan Zhang