[Eeglablist] Question about ASR and 'availableRAM_GB' parameter

[Velu Prabhakar Kumaravel]

Hi Nastassja,

When you do not fix the 'availableRAM_GB' parameter, everytime you run ASR,
you might see slightly different results. This is because the ASR algorithm
computes the available RAM size to avoid the memory issues during the
computation. For example, if ASR finds only 1 GB of RAM, it will calibrate
its parameters to process the data within that limit.

So, it is always recommended to fix the RAM so that you avoid the
replicability issues. Please note that the available physical memory
changes depending on the current workload.

I am not an expert in ICA but I imagine that when you specified the RAM
size, the algorithm did not remove the artifactual samples (eye blinks) but
instead, reconstructed them. Therefore, you didn't find the related IC
components. The best way to confirm this is to make note of the RAM size
when you do not specify a value for  'availableRAM_GB' to verify if it's
the same value you specified in the other case.

Best,
Velu Prabhakar Kumaravel, PhD Student
Center for Mind/Brain Sciences,
University of Trento, Italy


On Thu, 23 Jan 2020 at 08:35, Nastassja Lopes Fischer <fasnlf at nus.edu.sg>
wrote:

> Hi all!
>
> I am using the clean_rawdata plugin in children EEG data and it's been
> very useful to clean the signal. However, if I follow Makoto's
> recommendations written here (
> https://sccn.ucsd.edu/wiki/Artifact_Subspace_Reconstruction_(ASR)#Refe)
> about setting up a value for the 'availableRAM_GB' parameter, it gives me a
> "bad" ICA decomposition afterwards (i.e. ICA it is not able to isolate
> blink ICs, for instance). When I do not specify a value for
> 'availableRAM_GB', it improves ICA's results. Does it happen with anyone
> else? Would anyone know why it might occur? How critical is to not set up a
> value for 'availableRAM_GB'?
>
> Thanks in advance for the help!
>
> Best,
>
> Nastassja
>
> ________________________________
>
> Important: This email is confidential and may be privileged. If you are
> not the intended recipient, please delete it and notify us immediately; you
> should not copy or use it for any purpose, nor disclose its contents to any
> other person. Thank you.
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
>