Hi,<div><br></div><div>I'm running ICA on a dataset with 128 electrodes. Because I have RAM constraints (8GB), I wanted to know the amount of data required for a<i> 'good'</i> ICA. </div><div><br></div><div>I know the questions I'm asking don't have precise answers, but a rough guess by anyone with some experience in EEG analysis would be better than my own guess.</div>
<div><br></div><div>The wiki on 'decomposing ICA data' (<a href="http://sccn.ucsd.edu/wiki/Chapter_09:_Decomposing_Data_Using_ICA">http://sccn.ucsd.edu/wiki/Chapter_09:_Decomposing_Data_Using_ICA</a>) has the following to say about this issue:</div>
<div><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; ">"</span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; ">As a general rule, finding</span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "> </span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "><i>N</i></span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; ">stable components (from N-channel data) typically requires</span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "> </span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "><i>more than</i></span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "> </span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "><i>kN^2</i></span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "> </span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; ">data sample points (at each channel), where N^2 is the number of weights in the unmixing matrix that ICA is trying to learn and</span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "> </span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "><i>k</i></span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "> </span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; ">is a multiplier. In our experience, the value of</span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "> </span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "><i>k</i></span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "> </span><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; ">increases as the number of channels increases. In our example using 32 channels, we have 30800 data points, giving 30800/32^2 = 30 pts/weight points. However, to find 256 components, it appears that even 30 points per weight is not enough data."</span></div>
<div><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; "><br></span></div><div><span class="Apple-style-span" style="font-family: sans-serif; font-size: 13px; line-height: 19px; ">what would the rough value of <i>k</i> be for 128 channels?</span></div>
<div><font class="Apple-style-span" face="sans-serif"><span class="Apple-style-span" style="line-height: 19px;"><br></span></font></div><div><font class="Apple-style-span" face="sans-serif"><span class="Apple-style-span" style="line-height: 19px;">Regards,</span></font></div>
<div><font class="Apple-style-span" face="sans-serif"><span class="Apple-style-span" style="line-height: 19px;">Nishant Seth</span></font></div><div><font class="Apple-style-span" face="sans-serif"><span class="Apple-style-span" style="line-height: 19px;"><br>
</span></font></div><div><font class="Apple-style-span" face="sans-serif"><span class="Apple-style-span" style="line-height: 19px;">- - - </span></font><span class="Apple-style-span" style="font-family: sans-serif; line-height: 19px; ">- - - </span><span class="Apple-style-span" style="font-family: sans-serif; line-height: 19px; ">- - - </span><span class="Apple-style-span" style="font-family: sans-serif; line-height: 19px; ">- - - </span><span class="Apple-style-span" style="font-family: sans-serif; line-height: 19px; ">- - -</span></div>
<div><font class="Apple-style-span" face="sans-serif"><span class="Apple-style-span" style="line-height: 19px;">Research Associate,</span></font></div><div><font class="Apple-style-span" face="sans-serif"><span class="Apple-style-span" style="line-height: 19px;">Multimedia Lab, I.I.T. Delhi</span></font></div>
<div><font class="Apple-style-span" face="sans-serif"><span class="Apple-style-span" style="line-height: 19px;">New Delhi, India</span></font></div>