In the last few years an important family of methods for
single-channel signal separation has been developed using
tools from time-frequency analysis.
Given a mixture of signals f=∑i fi, the task is to
estimate the components fi using specific assumptions on their
time-frequency or statistical characteristics. A well known
strategy, denominated independent subspace
analysis (ISA), is to reduce the embedding dimension of
the time-frequency representation of f, prior to the application
of independent component analysis (ICA). In these methods,
a standard strategy for dimensionality reduction
is principal component analysis (PCA), but also nonlinear
methods have recently been proposed. This talk provides insight into basic principles of ISA, where we compare
different dimensionality reduction methods for single channel
signal separation in the context of ISA. Our focus is on signals
with transitory components, and the objective is to detect the
locations in time where each individual signal fi is activated.
This talk is based on joint work with Mijail Guillemard and Armin Iske.