0
$\begingroup$

This question arose from the Wikipedia page (https://en.wikipedia.org/wiki/Common_spatial_pattern). It says in the formula $$ {{w}}={\arg \max }_{{\mathbf {w}}}{\frac {\left\|{\mathbf {wX}}_{1}\right\|^{2}}{\left\|{\mathbf {wX}}_{2}\right\|^{2}}}$$ that we need to maximize the ratio between variances. So both the enumerator and denominator are variances. But X1 and X2 are simply matrices and, when multiplied by row vectors w become vectors. So why does this article say that the denominator and enumerator are variances? The way I see it, there should be something like E[(X-E[X])] as is explained in this Wikipedia article (https://en.wikipedia.org/wiki/Covariance_matrix#Conflicting_nomenclatures_and_notations) about variance $${ {E} \left[(\mathbf {X} -\mathrm {E} [\mathbf {X} ])(\mathbf {X} -\mathrm {E} [\mathbf {X} ])^{\rm {T}}\right]}$$.

So my problem is I don't see where this first formula above 'hides' variance. Is it some simplified representation? Please help.

$\endgroup$
0
$\begingroup$

I think I figured it out. If we ignore the w factors, bothe fenominator and the enumerator are variances. Wikipedia failed to mention that the signals are centered, i.e. have zero means. The (1/1-N) factors that are part of the variance formula cancel out since both signals have the same number of samples.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.