What is whitening ICA?

What is whitening ICA?

Whitening. Before applying the ICA algorithm, we must first “whiten” our signal. To “whiten” a given signal means that we transform it in such a way that potential correlations between its components are removed (covariance equal to 0) and the variance of each component is equal to 1.

What does it mean to whiten data?

Whitening, or sphering, data means that we want to transform it to have a covariance matrix that is the identity matrix — 1 in the diagonal and 0 for the other cells. It is called whitening in reference to white noise.

What is whitening in image?

Whitening or sphering data means that we want to transform it in a way to have a covariance matrix that is the identity matrix (1 in the diagonal and 0 for the other cells; more details on the identity matrix). It is called whitening in reference to white noise.

Which is more important, PCA or ICA?

PCA removes correlations, but not higher order dependence ICA removes correlations andhigher order dependence PCA: some components are more important than others (recall eigenvalues) ICA: all components are equally important PCA: vectors are orthogonal (recall eigenvectors of covariance matrix)

Can you remove an epoch from an ICA algorithm?

Removing data samples in the form of entire epochs will have an effect on the results of ICA algorithms. Especially if those epochs contain large, unique movement artifacts. ICA algorithms will typically place unique movement artifacts in single components.

How are recall eigenvalues different from Ica and PCA?

Differences between ICA and PCA. PCA removes correlations, but not higher order dependence ICA removes correlations and higher order dependence PCA: some components are more important than others (recall eigenvalues) ICA: all components are equally important. PCA: vectors are orthogonal (recall eigenvectors of covariance matrix)

Is it better to apply ICA on whole EEG data?

ICA algorithms will typically place unique movement artifacts in single components. It would probably be best to submit all epochs to the algorithm unless there are some egregious movement artifacts in the recording. No EEG will be able to be recovered during periods of large movement artifact.