Andreas Ziehe

An approach to blind source separation based on temporal structure of speech signals

Abstract In this paper, we introduce a new technique for blind source separation of speech signals. We focus on the temporal structure of the signals. The idea is to apply the decorrelation method proposed by Molgedey and Schuster in the time–frequency domain. Since we are applying separation algorithm on each frequency separately, we have to solve the amplitude and permutation ambiguity properly to reconstruct the separated signals. For solving the amplitude ambiguity, we use the matrix inversion and for the permutation ambiguity, we introduce a method based on the temporal structure of speech signals. We show some results of experiments with both artificially controlled data and speech data recorded in the real environment.

Robustly estimating the flow direction of information in complex physical systems

We propose a new measure (phase-slope index) to estimate the direction of information flux in multivariate time series. This measure (a) is insensitive to mixtures of independent sources, (b) gives meaningful results even if the phase spectrum is not linear, and (c) properly weights contributions from different frequencies. These properties are shown in extended simulations and contrasted to Granger causality which yields highly significant false detections for mixtures of independent sources. An application to electroencephalography data (eyes-closed condition) reveals a clear front-to-back information flow.

TDSEP — an efficient algorithm for blind separation using time structure

An algorithm for blind source separation based on several time-delayed second order correlation matrices is proposed. The technique to construct the unmixing matrix employs first a whitening step and then an approximate simultaneous diagonalisation of several time-delayed second order correlation matrices. Its efficiency and stability are demonstrated for linear artificial mixtures with 17 sources.

Artifact reduction in magnetoneurography based on time-delayed second-order correlations

Artifacts in magnetoneurography data due to endogenous biological noise sources, like the cardiac signal, can be four orders of magnitude higher than the signal of interest. Therefore, it is important to establish effective artifact reduction methods. We propose a blind source separation algorithm using only second-order temporal correlations for cleaning biomagnetic measurements of evoked responses in the peripheral nervous system. The algorithm showed its efficiency by eliminating disturbances originating from biological and technical noise sources and successfully extracting the signal of interest. This yields a significant improvement of the neuro-magnetic source analysis.

Kernel-based nonlinear blind source separation

We propose kTDSEP, a kernel-based algorithm for nonlinear blind source separation (BSS). It combines complementary research fields: kernel feature spaces and BSS using temporal information. This yields an efficient algorithm for nonlinear BSS with invertible nonlinearity. Key assumptions are that the kernel feature space is chosen rich enough to approximate the nonlinearity and that signals of interest contain temporal information. Both assumptions are fulfilled for a wide set of real-world applications. The algorithm works as follows: First, the data are (implicitly) mapped to a high (possibly infinite)dimensional kernel feature space. In practice, however, the data form a smaller submanifold in feature spaceeven smaller than the number of training data pointsa fact that has already been used by, for example, reduced set techniques for support vector machines. We propose to adapt to this effective dimension as a preprocessing step and to construct an orthonormal basis of this submanifold. The latter dimension-reduction step is essential for making the subsequent application of BSS methods computationally and numerically tractable. In the reduced space, we use a BSS algorithm that is based on second-order temporal decorrelation. Finally, we propose a selection procedure to obtain the original sources from the extracted nonlinear components automatically. Experiments demonstrate the excellent performance and efficiency of our kTDSEP algorithm for several problems of nonlinear BSS and for more than two sources.

Combining sparsity and rotational invariance in EEG/MEG source reconstruction.

We introduce Focal Vector Field Reconstruction (FVR), a novel technique for the inverse imaging of vector fields. The method was designed to simultaneously achieve two goals: a) invariance with respect to the orientation of the coordinate system, and b) a preference for sparsity of the solutions and their spatial derivatives. This was achieved by defining the regulating penalty function, which renders the solutions unique, as a global l(1)-norm of local l(2)-norms. We show that the method can be successfully used for solving the EEG inverse problem. In the joint localization of 2-3 simulated dipoles, FVR always reliably recovers the true sources. The competing methods have limitations in distinguishing close sources because their estimates are either too smooth (LORETA, Minimum l(1)-norm) or too scattered (Minimum l(2)-norm). In both noiseless and noisy simulations, FVR has the smallest localization error according to the Earth Movers Distance (EMD), which is introduced here as a meaningful measure to compare arbitrary source distributions. We also apply the method to the simultaneous localization of left and right somatosensory N20 generators from real EEG recordings. Compared to its peers FVR was the only method that delivered correct location of the source in the somatosensory area of each hemisphere in accordance with neurophysiological prior knowledge.

A resampling approach to estimate the stability of one-dimensional or multidimensional independent components

When applying unsupervised learning techniques in biomedical data analysis, a key question is whether the estimated parameters of the studied system are reliable. In other words, can we assess the quality of the result produced by our learning technique? We propose resampling methods to tackle this question and illustrate their usefulness for blind-source separation (BSS). We demonstrate that our proposed reliability estimation can be used to discover stable one-dimensional or multidimensional independent components, to choose the appropriate BSS-model, to enhance significantly the separation performance, and, most importantly, to flag components that carry physical meaning. Application to different biomedical testbed data sets (magnetoencephalography (MEG)/electrocardiography (ECG)-recordings) underline the usefulness of our approach.

Combining time-delayed decorrelation and ICA: towards solving the cocktail party problem

We present methods to separate blindly mixed signals recorded in a room. The learning algorithm is based on the information maximization in a single layer neural network. We focus on the implementation of the learning algorithm and on issues that arise when separating speakers in room recordings. We used an infomax approach in a feedforward neural network implemented in the frequency domain using the polynomial filter matrix algebra technique. A fast convergence speed was achieved by using a time-delayed decorrelation method as a preprocessing step. Under minimum-phase mixing conditions this preprocessing step was sufficient for the separation of signals. These methods successfully separated a recorded voice with music in the background (cocktail party problem). Finally, we discuss problems that arise in real world recordings and their potential solutions.

On-line learning in changing environments with applications in supervised and unsupervised learning

An adaptive on-line algorithm extending the learning of learning idea is proposed and theoretically motivated. Relying only on gradient flow information it can be applied to learning continuous functions or distributions, even when no explicit loss function is given and the Hessian is not available. The framework is applied for unsupervised and supervised learning. Its efficiency is demonstrated for drifting and switching non-stationary blind separation tasks of acoustic signals. Furthermore applications to classification (US postal service data set) and time-series prediction in changing environments are presented.

BLIND SOURCE SEPARATION TECHNIQUES FOR DECOMPOSING EVENT-RELATED BRAIN SIGNALS

Recently blind source separation (BSS) methods have been highly successful when applied to biomedical data. This paper reviews the concept of BSS and demonstrates its usefulness in the context of event-related MEG measurements. In a first experiment we apply BSS to artifact identification of raw MEG data and discuss how the quality of the resulting independent component projections can be evaluated. The second part of our study considers averaged data of event-related magnetic fields. Here, it is particularly important to monitor and thus avoid possible overfitting due to limited sample size. A stability assessment of the BSS decomposition allows to solve this task and an additional grouping of the BSS components reveals interesting structure, that could ultimately be used for gaining a better physiological modeling of the data.