Blind Source Separation

The implantation of Blind Source Separation.

Usage

Independent Vector Analysis

from independent_vector_analysis import IndependentVectorAnalysis

iva = IndependentVectorAnalysis(21)
iva.fit(X)
y = iva.fit_transform(X)

Raw data. Independent Vector. Computing every autocorrection of Independent Vector, if the autocorrection is below 0.9, set it zero. After that, reconstruct data.

Online Independent Component Analysis

from ORICA import OnlineRecursiveIndependentComponentAnalysis
orica = OnlineRecursiveIndependentComponentAnalysis(num_channels=32, sfreq=200, blockSize=200)
for i in range(0, data.shape[-1], 32):
    block = data[:, i:i + 32]
    ica_data = orica.push(block)
    c = np.append(c, ica_data, axis=1)
c = np.array(c)[:, 1:]