A. Ziehe

Publications of A. Ziehe

• Robustly estimating the flow direction of information in complex physical systems

  Authors: G. Nolte, A. Ziehe, V.V. Nikulin, A. Schlögl, N. Krämer, T. Brismar, K.R. Müller

  Arxiv preprint arXiv:0712.2352. 01/2007;

• Analyzing Coupled Brain Sources: Distinguishing True from Spurious Interaction

  Authors: G. Nolte, A. Ziehe, F. Meinecke, K.R. Müller

  01/2006

• A fast algorithm for joint diagonalization with non-orthogonal transformations and its application to blind source separation

  Authors: A. Ziehe, P. Laskov, G. Nolte, K.R. Müller

  Journal of Machine Learning Research. 01/2004; 5:777--800.

• Estimating the Reliability of ICA

  Authors: F. Meinecke, A. Ziehe, M. Kawanabe

  10/2003;

  When applying unsupervised learning techniques like ICA or temporal decorrelation, a key question is whether the discovered projections are reliable. In other words: can we give error bars or can weWhen applying unsupervised learning techniques like ICA or temporal decorrelation, a key question is whether the discovered projections are reliable. In other words: can we give error bars or can we assess the quality of our separation? We use resampling methods to tackle these questions and show experimentally that our proposed variance estimations are strongly correlated to the separation error. We demonstrate that this reliability estimation can be used to choose the appropriate ICA-model, to enhance significantly the separation performance, and, most important, to mark the components that have a actual physical meaning. Application to 49-channel-data from an magnetoencephalography (MEG) experiment underlines the usefulness of our approach.

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

  Authors: F. Meinecke, A. Ziehe, M. Kawanabe, K.-R. Muller

  Biomedical Engineering, IEEE Transactions on. 01/2003;

  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 theWhen 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.

• Estimating the Reliability of ICA Projections

  Authors: F. Meinecke, A. Ziehe, M. Kawanabe

  04/2002;

  When applying unsupervised learning techniques like ICA or temporal decorrelation, a key question is whether the discovered projections are reliable. In other words: can we give error bars or can weWhen applying unsupervised learning techniques like ICA or temporal decorrelation, a key question is whether the discovered projections are reliable. In other words: can we give error bars or can we assess the quality of our separation? We use resampling methods to tackle these questions and show experimentally that our proposed variance estimations are strongly correlated to the separation error. We demonstrate that this reliability estimation can be used to choose the appropriate ICA-model, to enhance significantly the separation performance, and, most important, to mark the components that have a actual physical meaning. Application to 49-channel-data from an magnetoencephalography (MEG) experiment underlines the usefulness of our approach.

• A comparison of ICA-based artifact reduction methods for MEG

  Authors: A. Ziehe, G Nolte, K. -r. Mller, G Curio

  12/2001;

  Introduction In the analysis of MEG data one often faces the problem that noise from biological or technical origins (e.g. alpha activity or interference from the power line, respectively) isIntroduction In the analysis of MEG data one often faces the problem that noise from biological or technical origins (e.g. alpha activity or interference from the power line, respectively) is corrupting the measurements. We present a case study where we analyze the effects of artifact removal for a well-known experimental setting: measurements of somatosensory evoked fields (SEF, N20). We compare a classical signal processing approach to the recently developed independent component analysis (ICA) technology [9, 11]. The specific data set studied is an attractive testbed since the signal of interest (N20) is relatively strong, but contaminated by a 150 Hz component due to power line interference. 2 Methods 2.1 Data The right median nerve was stimulated electrically over 12000 epochs, while the magnetic field above the contralateral somatosensory cortex was measured by the Berlin 49 channel planar gradiometer system placed in a magnetically shielded room. A sampling rate of 2 kHz and

• Noise robust estimates of correlation dimension and K2 entropy.

  Authors: G Nolte, A. Ziehe, K.-R. Müller

  Physical review. E, Statistical, nonlinear, and soft matter physics. 08/2001; 64(1 Pt 2):016112.

  Using Gaussian kernels to define the correlation sum we derive simple formulas that correct the noise bias in estimates of the correlation dimension and K2 entropy of chaotic time series. TheUsing Gaussian kernels to define the correlation sum we derive simple formulas that correct the noise bias in estimates of the correlation dimension and K2 entropy of chaotic time series. The corrections are only based on the difference of correlation dimensions for adjacent embedding dimensions and hence preserve the full functional dependencies on both the scale parameter and embedding dimension. It is shown theoretically that the estimates, which are derived for additive white Gaussian noise, are also robust for moderately colored noise. Simulations underline the usefulness of the proposed correction schemes. It is demonstrated that the method gives satisfactory results also for non-Gaussian and dynamical noise.

  Download

• Independent component analysis of noninvasively recorded cortical magnetic DC-fields in humans.

  Authors: G Wübbeler, A. Ziehe, B M Mackert, K.-R. Müller, L Trahms, G Curio

  IEEE transactions on bio-medical engineering. 06/2000; 47(5):594-9.

  We apply a recently developed multivariate statistical data analysis technique--so called blind source separation (BSS) by independent component analysis--to process magnetoencephalogram recordingsWe apply a recently developed multivariate statistical data analysis technique--so called blind source separation (BSS) by independent component analysis--to process magnetoencephalogram recordings of near-dc fields. The extraction of near-dc fields from MEG recordings has great relevance for medical applications since slowly varying dc-phenomena have been found, e.g., in cerebral anoxia and spreading depression in animals. Comparing several BSS approaches, it turns out that an algorithm based on temporal decorrelation successfully extracted a dc-component which was induced in the auditory cortex by presentation of music. The task is challenging because of the limited amount of available data and the corruption by outliers, which makes it an interesting real-world testbed for studying the robustness of ICA methods.

• Independent component analysis of noninvasively recorded cortical magnetic DC-fields in humans

  Authors: G. Wubbeler, A. Ziehe, B.-M. Mackert, K.-R. Muller, L. Trahms, C. Curio

  Biomedical Engineering, IEEE Transactions on. 06/2000;

  We apply a recently developed multivariate statistical data analysis technique-so called blind source separation (BSS) by independent component analysis-to process magnetoencephalogram recordings ofWe apply a recently developed multivariate statistical data analysis technique-so called blind source separation (BSS) by independent component analysis-to process magnetoencephalogram recordings of near-DC fields. The extraction of near-DC fields from MEG recordings has great relevance for medical applications since slowly varying DC-phenomena have been found, e.g., in cerebral anoxia and spreading depression in animals. Comparing several BSS approaches, it turns out that an algorithm based on temporal decorrelation successfully extracted a DC-component which was induced in the auditory cortex by presentation of music. The task is challenging because of the limited amount of available data and the corruption by outliers, which makes it an interesting real-world testbed for studying the robustness of ICA methods.

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

  Authors: A. Ziehe, K.-R. Müller, G Nolte, B M Mackert, G Curio

  IEEE transactions on bio-medical engineering. 02/2000; 47(1):75-87.

  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 isArtifacts 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.

• Decomposition algorithms for analysing brain signals

  Authors: K.-R. Muller, J. Kohlmorgen, A. Ziehe, B. Blankertz

  Adaptive Systems for Signal Processing, Communications, and Control Symposium 2000. AS-SPCC. The IEEE 2000; 02/2000

  Analyzing biomedical data-e.g. from the brain-we encounter fundamental problems that lie largely in the fields of signal processing and machine learning. The current paper presents at first a methodAnalyzing biomedical data-e.g. from the brain-we encounter fundamental problems that lie largely in the fields of signal processing and machine learning. The current paper presents at first a method to deal with non-stationary signals, subsequently the signal processing technique of independent component analysis (ICA) is reviewed. We use EEG recordings of continuous auditory perception as illustration for the discussed algorithms

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

  Authors: A. Ziehe, K.-R. Muller, G. Nolte, B.-M. Mackert, G. Curio

  Biomedical Engineering, IEEE Transactions on. 02/2000;

  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 isArtifacts 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

• Separation of post-nonlinear mixtures using ACE and temporal decorrelation

  Authors: A. Ziehe, M. Kawanabe, S. Harmeling, K.-R. Müller

  Fraunhofer FIRST.