Background

This section explains the ideas behind pyAMICA from the ground up, for readers new to independent component analysis.

• What is ICA? — the blind source separation problem, the linear mixing model, and why statistical independence and non-Gaussianity make it solvable.
• What is AMICA? — how AMICA extends ICA with adaptive source densities (mixtures of generalized Gaussians) and multiple ICA models.
• How AMICA works — the log-likelihood objective and the expectation-maximization algorithm (natural-gradient and Newton updates) that fits it.

If you just want to run a decomposition, start with Getting Started instead.