x x2 H11 H y y2 H22 H

Size: px
Start display at page:

Download "x x2 H11 H y y2 H22 H"

Transcription

1 H H2 5 5 x 5 5 x H H y.5.5 y2 Figure : Mixing coef.: H [k], H 2 [k], H 2 [k], H 22 [k]. Figure 3: Observations and recovered sources G G PEFOMANCE INDEX G G ITEATIONS Figure 2: Overall coef.: G [k], G 2 [k], G 2 [k], G 22 [k]. Figure 4: Performance Index versus iterations. Cichocki, for their useful discussion during this period. This work has also been supported by the Grants: CI- CYT (TIC96-5-C-8), CICYT (TIC96-5-C- 2) and Xunta de Galicia (XUGA 52A96). 9. EFEENCES [] L. B. Almeida and F. M. Silva. Adaptive decorrelation. Articial Neural Networks (Elsevier), 2:49{56, 992. [2] S. Amari and J. F. Cardoso. Blind source separationsemiparametric statistical approach. IEEE Transactions on Signal Processing, 45():2692{2697, 997. [3] S. Amari, S. Douglas, A. Cichocki, and H. Yang. Multichannel blind deconvolution and equalization using the natural gradient. In IEEE Workshop on Wireless Communication, Paris, pages {4, April 997. [4] S.-I. Amari, T.-P. Chen, and A. Cichocki. Stability analysis of learning algorithms for blind source separation. Neural Networks, (8):345{35, 997. [5] A. Bell and T. Sejnowski. An information maximization approach to blind separation and blind deconvolution. Neural Computation, (7,6):29{59, 996. [6] X.-. Cao and. Wen Liu. General approach to blind source separation. IEEE Transactions on Signal Processing, 44(3):562{57, Mar [7] J. Cardoso and B. Laheld. Equivariant adaptive source separation. IEEE Transactions on Signal Processing, 44(2):37{33, Dec [8] A. Cichocki,. Bogner, L. Moszczynski, and K. Pope. Modied H-J algorithms for blind separation of sources. Digital Signal Processing, 7(2):8{93, 997. [9] A. Cichocki and. Unbehauen. obust neural networks with on-line learning for blind identication and blind separation of sources. IEEE Transactions on Circuits and Systems-I, 43():894{96, 996. [] A. Cichocki,. Unbehauen, and E. ummert. obust learning algorithm for blind separation of signals. Electronics Letters, 3(7):386{387, August 994. [] P. Comon. Independent component analysis, a new concept? Signal Processing, (36):287{34, 994. [2] S. Cruces, A. Cichocki, and L. Castedo. An unied perspective of blind source separation adaptive algorithms. In proceedings of the Learning'98 congress (to appear), Madrid, Spain, Sept [3] S. Douglas and A. Cichocki. Neural networks for blind decorrelation of signals. IEEE Transactions on Signal Processing, 45():2829{2842, Nov 997. [4]. Fletcher. Practical Methods of Optimization. John Willey and Sons, second edition, 987. [5] S. Haykin. Blind Deconvolution. Prentice Hall, 994. [6] C. Jutten and J. Herault. Blind separation of sources, part I: An adaptive algorithm based on neuromimetic architecture. Signal Processing, (24):{, 99. [7] J. Karhunen, E. Oja, L. Wang,. Vigario, and J. Koutsensalo. A class of neural networks for independent component analysis. IEEE Transactions on Neural Networks, 8(3):486{53, May 97.

2 From the denition of the PSD matrix we see that ks (!) ; Ik 2 +ks (!)k 2 + P k k [k]k 2 Close to convergence, the following approximation Pk k [k]k 2 maxfjdiag( [])jg holds true and choosing suciently small so that P + k maxfjdiag( [k])jg P + k k [k]k 2 we arrive at the practical step-size for the convolutive mixtures separation algorithm given by = + P k maxfjdiag( [k])jg (28) 5. STABILITY ANALYSIS In this section we perform a stability analysis of the II method and provide the necessary and sucient conditions that the non-linearities must satisfy in order to ensure that the separating solutions are stable points. At the separating solution the sources s[n] should be properly scaled in order that [] = I holds true. To simplify the notation we will dene f i = f(s i [n]), g i = g(s i [n]), s i = s i [n], i[n] and i[n], for all i = N. It can be shown (see [2]) that (for real sources) the necessary and sucient conditions that ensure asymptotic stability of the instantaneous and convolutive II methods are E[fi s ig i ]+E[f i s i gi] > (29) E[fi]E[s j g j ]+E[fj]E[s i g i ] > (3) E[fi ]E[f j ]E[s ig i ]E[s j g j ] >E[gi ]E[g j ]E[s if i ]E[s j f j ](3) for all i jj i6=j. The rst two conditions can be easily satised by the use of strictly monotonically increasing odd functions. The third condition is not as critical since, as suggested in [4], it can be forced by just permuting the order of the functions f() andg(). Similar results have been obtained by Amari et al. [4] and Cardoso [7] when g(y) = y. Our results, however, are valid for the general case of two non-linear functions. ecently, Amari and Cardoso have shown in [2] that the best local estimating function (f(y)g T (y)) is obtained when one of the functions f() org() is linear, i.e., when f(y) = y or g(y) = y. Nevertheless, they also point out that their results characterize only the local asymptotic behavior while other estimating functions, with two non-linearities, may have preferred global properties and are more robust for noisy sensor signals. 6. COMPUTE SIMULATIONS Computer simulations were carried out to illustrate the performance of the II method for the separation of convolutive mixtures. We considered two i.i.d. 6- QAM sources and the following mixing system H[k] = [:z 3 +:2z 2 +:6z + ;:z 3 +:2z 2 ; :4z +:5 ; :4z ; +:2z ;2 ; :z ;3 :z 3 ; :3z 2 +:5z ; :6+ :5z ; ;:3z ;2 +:z ;3 +:6z ; +:2z ;2 +:z ;3 ] (see Figure ). The separating system is a two-input two-output system with four FI lters having 2 taps each. To adapt the separating system we used the algorithm given by eq. (26) and (28) in its batch version overadatawindow of samples. The chosen parameters were =,f(y) =jyj 2 y and g(y) =y. We dened the performance index P index of the separation system in terms of the overall response G[n] as P index = X M X i= + MX j= j= k NX X i= jg ij [k]j 2 max l m fjg il[m]j 2 g ; k A! jg ij [k]j 2 max l m fjg lj[m]j 2 g ; (32) Figure 4 illustrates the evolution of the performance index through the iterations, where it is clearly seen that convergence is achieved after twenty iterations. Finally, Figure 3 shows the observations of the two mixed sources and the resulting outputs after convergence. 7. CONCLUSIONS A new way to derive the generalized learning rule proposed in [] has been presented. The new approach is termed Iterative Inversion (II) method because it is an iterative procedure to invert a non-linear correlation matrix of the outputs. Although initially presented for instantaneous mixtures, the method is also extended to consider the more realistic case of convolutive mixtures. It is also noted that many existing algorithms can be interpreted as particular cases of the II method thus providing a unied perspective to BSS. Finally, a stability analysis shows that choosing strictly monotically increasing odd non-linear functions ensures local convergence to the separation solution. 8. ACKNOWLEDGMENTS Part of this workistheresultofastay at the Brain Science Institute (IKEN), supported by the Frontier esearch Program of Japan. The rst author wishes to thank the members of the Brain Information Processing esearch Group and especially Profs. Amari and

3 the other hand, the second matrix W b will be selected to diagonalize the skew-hermitian part of and towards this aim the on-line version of the algorithm (5) can be used. However, in order to merge (5) and (22) into a single recursion for the overall separating system W we need that both adaptations be orthogonal at a rst order. Since the matrix yy H ; I is always Hermitian, to orthogonalize the second adaptation at a rst order we can replace f(y)g H (y) ; I by its projection onto the space of skew-hermitian matrices and the following algorithm results W (n+) b = I ; f(y)g H (y) ; g(y)f H (y) +jf H (y)g(y)j W b Combining both adaptation rules as W (n+) = W (n+) b and keeping only the terms of the rst order W (n+) a expansion in (which isavalid approximation for small enough) we arriveat W (n+) = yy I H ; I ; +jy H yj ; f(y)g H (y) ; g(y)f H (y) +jf H (y)g(y)j W which is a normalized version of the family of generalized EASI algorithms [7, 7]. -Non-linear PCA algorithm: Imposing the constraint that the separation matrix W be unitary, we can derive the non-linear PCA algorithm developed by Oja and Karhunen [7] from the II method. First, let us redene the nonlinear correlation matrix as = E[f(y)gH (y)] + I where f(y) is an odd function and g(y) =f(y) ; y. Substituting now the matrix by its stochastic approximation in the II algorithm (5) we arrive at W (n+) = W ; ; I W (23) = W ; f(y) ; f H (y) ; y H W Taking into account that W is a unitary matrix, i.e., W H W = I, we can rewrite (23) as W (n+) = W + f(y) x H ; f H (y)w which is Karhunen-Oja non-linear PCA algorithm [7]. 4. CONVOLUTIVE MIXTUES In this section we will extend the ideas presented in the previous section to the case of convolutive mixtures of temporally i.i.d. sources. In this case, we propose the diagonalization of the following nonlinear correlation matrix sequence [k] = E[f(y[n])gH (y[n ; k])] = [k] I (24) where f() andg() are suitably chosen non-linear functions. Let us dene the nonlinear Power Spectral Density (PSD) matrix S (!) as the Discrete Time Fourier Transform (DTFT) of the correlation matrix [k]. Similarly,wedeneU (!) =DTFTfW [k]g. Therefore, in the frequency domain equation (24) is equivalent tohaving a nonlinear PSD matrix equal to the identity matrix for all the frequencies, i.e. S (!) = I 8!. Moreover, we can dene a new non-linear PSD matrix S Fg (!) =U; (!) sothat S S (!) ; (!) =I ) U (!) =S Fg (!) (25) It is interesting to note that, for each frequency bin!, we have a decoupled instantaneous mixture problem under the assumption that the separation lter has an innite number of coecients. Then, we can use to calculate the separation matrix at each frequency the II method presented in Section 2 resulting in the following frequency domain algorithm U (n+) (!)=U (!) ; S (!) ; I U (!) =(+ )U (!) ; S (!)U (!) Working in the frequency domain has the problem that for each frequency bin there is an independent separation algorithm that may yield to the sources with permuted orders. This limitation can be overcome if we compute the inverse DTFT of the algorithm (26) to express it in the time domain W (n+) [k] = ( + )W [k] ; [k] (26) [k] = [k] W [k] = X m= m=; [m] W [k ; m] The stochastic version of the algorithm (26) is obtained by means of the approximation [m] f(y[n]) g H (y[n;m]). Following a similar procedure to section 3, it is possible to chose the non-linearities so as to extend the existing algorithms for instantaneous mixtures to convolutive mixtures. To complete the denition of the iteration we still have to determine the step-size. In order to ensure convergence for all the frequencies should satisfy < ks (!) ; Ik 2 8! (27)

4 It should be noted that the derived above algorithm (4) is extended batch version of the family of adaptive algorithms proposed originally by Cichocki et al. in [, 9, 8] for blind source separation. This algorithm exhibits the equivariant property in the absence of noise On-line and batch adaptations The adaptive implementation of the II algorithm (5) admits both on-line and batch versions. On-line adaptation: the correlation matrix in (5) is replaced by its single sample estimation at time n, i.e., f(y[n])gh (y[n]) (6) Batch adaptation: if we assume that the stationarity and ergodicity properties hold on a block of observations of L samples we can replace in (5) the statistical average by the temporal average L L; X k= f(y[k])g H (y[k]) (7) 2.3. A practical step-size adaptation In order to provide numerical stability of the algorithms and to ensure high convergence speed it is importantto choose possibly close to optimal learning step. Taking into account that for the separation algorithm (5) we needed to ensure the condition (3) and since k ; Ik 2 +k k 2 we propose to estimate the learning step as = +k k 2 : (8) This means that for the on-line adaptation resulting in = k k 2 = jg H (y[n]) f(y[n])j +jg H (y[n]) f(y[n])j (9) In the batch case, however, taking into account that the algorithm diagonalizes, a practical estimate of the spectral radius of the correlation matrix is given by k k 2 maxfjdiag( )jg This approximation may not be true at points far from convergence. However, if is small enough, the condition for the expansion of the inverse (3) is still true. Therefore, we propose to use = +maxfjdiag( )jg (2) 3. ELATION WITH EXISTING ALGOITHMS Several existing algorithms for BSS can be derived from the II method (5): the decorrelation algorithm of Almeida [], the natural gradient algorithm for ICA [3, 2], the non-linear PCA algorithm [7] (under the requirement ofhaving an unitary separation matrix) and the normalized and generalized EASI algorithms [7, 7]. In fact, all these models can be shown to be consistent with (5) when one or more linear/non-linear correlation matrices are diagonalized. Below, we will show these relations for the on-line case. - Decorrelation algorithm: In the linear case f(y) =g(y) =y and the adaptation rule (5) reduces to the decorrelation algorithm proposed by Almeida [] (see also Douglas and Cichocki [3, 9]). W (n+) = W ; yy H ; I +jy H yj W (2) - Amari's Natural gradient algorithms for ICA: When only one of the two functions f(y) org(y) is non-linear the II algorithm converts into W (n+) = W ; f(y) y H ; I +jy H f(y)j W (22) or in its permuted form. Both are normalized versions of the Natural Gradient algorithms for ICA developed by Amari et al. [3, 2] and independently by Cardoso and Laheld [7]. - Generalized EASI algorithms: The family of generalized EASI algorithms was proposed by Karhunen et al. in [7] as an extension of the family of EASI algorithms derived by Cardoso and Laheld in [7]. We can derive its normalized version from the II method following a procedure similar to that described by Cardoso and Laheld in their article. Let us split the separation stage in two blocks connected in series, i.e., W = W b W a. The rst matrix W a will be selected to diagonalize the Hermitian matrix yy and will be adapted according to algorithm (2). On

5 derivation of the generalized and exible learning algorithm for instantaneous BSS is given. Section 3 shows that many existing algorithms for BSS can be derived as special cases of the II method. Section 4 extends the method to convolutive mixtures. Section 5 presents the conditions to ensure algorithm stability. Section 6 illustrates the performance through computer simulations and Section 7 is devoted to the conclusions. 2. INSTANTANEOUS MIXTUES 2.. Criterion According to the Darmois-Skitovich theorem [, 6], sources are recovered if and only if y is a vector of statistically independent signals. When this occurs, it is apparent that the nonlinear spatial correlation matrix (W) =E[f(y)g H (y)] (5) will be diagonal. Here f(y) and g(y) represent nonlinear functions of the outputs. Without loss of generality we will consider that this correlation matrix is equal to the identity at the optimum separation matrices W, i.e., (W )=I (6) Next, let us dene a new non-linear function of the outputs F(y) =W ; f(y) and a new spatial correlation matrix Fg (W) =E[F(y)g H (y)] = W ; (W) (7) This allows us to rewrite condition (6) as (W )=I ) W = ; Fg (W ) (8) Therefore, we mayinterpret that solving the BSS problem is equivalent toinverting the matrix Fg (W ). This inversion, however, cannot be carried out directly since we do not have access to F(y) andg(y) atthe optimum separating solution W.Instead,we will develop in the sequel an iterative procedure that overcomes this limitation. Let us denote W the separating matrix used in the nth iteration. According to the above discussion the best approximation to W at the nth iteration will be the inverse of Fg (W ), i.e., W (n+) = ; Fg (W ) (9) In order to avoid the computational complexity caused by matrix inversions, we propose that W (n+) be the solution to the following minimization problem arg min W ; (n+) =k Fg (W ) ; W ; (n+) k 2 F () where kk F denotes the Frobenius norm of a matrix. Let us resort now to a Newton-aphson recursive technique to solve the minimization problem (). Following an argumentation similar to the development of the Bussgang techniques for blind equalization [5] and to the continuation method [4], to extend the domain of convergence of the Newton-apshon method we willinterpret Fg (W ) as a rough estimate of the mixing matrix H at iteration n and, therefore, with a constant value. Computing the gradient and the Hessian of with respect to W ; (n+) it is obtained that Fg (W ) ; W ; = I. Therefore the apshon recursion that minimizes is W ; (n+) = W ; + ( Fg ; W; (n+) ) where to simplify notation Fg = Fg(W ). Dening = we can rewrite the above iteration as + W ; (n+) = ( ; ) W ; + Fg () Note from this equation that W ; (n+) is an estimate of an exponential windowed version of the spatial correlation matrix, and is in accordance with our interpretation of Fg as an estimate of the mixing system H. Moreover, it also provides some clues to select the nonlinearities F() and g() as those that provide a good estimate of H from the outputs vector y. Next, let us describe explicitly the recursion () in terms of W ; rather than W W (n+) = If is chosen in such away that < I + ; I ; W (2) k ; Ik 2 (3) where kk 2 denotes the 2-norm, we can express the inverse in (2) as a Taylor series expansion and make the following approximation I + ; I ; = X i= ; ; I i I ; ; I (4) Finally, substituting (4) into (2) we arrive at the following Iterative Newton-aphson method for the Inversion of Fg, the so-called Iterative Inversion (II) method W (n+) = W ; ; I W (5)

6 AN ITEATIVE INVESION METHOD FO BLIND SOUCE SEPAATION Sergio Cruces and Luis Castedo Universities of Seville and La Coru~na Camino de los Descubrimientos 492-Seville, Spain Andrzej Cichocki Laboratory for Open Information Systems Brain Science Institute, IKEN Wako-Shi, Saitama 35-, JAPAN ABSTACT In this paper we present an Iterative Inversion (II) approach to derive the generalized learning rule for Blind Source Separation (BSS) proposed in []. The approach consists in the diagonalization of a non-linear spatial correlation matrix of the outputs and is rst presented for instantaneous mixtures. It is shown how existing algorithms for BSS can be derived as particular cases of the resulting generalized rule. The II method is also extended to the separation of convolutive mixtures of temporally i.i.d. sources. Finally, necessary and sucient asymptotic stability conditions for the method to converge are given.. INTODUCTION The separation of mixtures of unknown signals arises in a large number of signal processing applications such as array processing, multiuser communications and voice restoration. This problem is known as Blind Source Separation (BSS) and it has been shown [] that can be solved if the sources are non-gaussian and statistically independent. Since the pioneering work of Jutten and Herault [6], a lot of novel, ecient and robust adaptive algorithms for blind source separation have been rigorously derived and their properties have been investigated. Algorithms are developed from a dierent points of view such ascontrast functions [], mutual information maximization [5], Kullback-Leibler minimization using natural gradient approach [3] and nonlinear principal component analysis [7]. In this paper we present a new approach, the Iterative Inversion (II) method, and show that most of the existing algorithms can be obtained as particular cases of this method. The BSS separation problem is typically formulated as follows. Assume that an array of sensors provides a vector of N observed signals x[n] =[x [n] x 2 [n] x N [n]] T that are mixtures of N random processes s i [n] i = 2 N termed sources. The exact probability density function of the sources is unknown: we only assume that they are complex-valued, zero-mean, non- Gaussian distributed and statistically independent. In the convolutive mixture case we consider that the sources samples are temporally independent and identically distributed (i.i.d.), and the observations are related to the sources as follows x[n]= X k=; H[k]s[n ; k] () where s[n] =[s [n] s 2 [n] s N [n]] T is the vector of sources and H[n] is the sequence of N N impulse response matrices corresponding to the mixing system. To recover the sources, the observations are processed by a Multiple Input Multiple Output (MIMO) system with memory to produce the outputs y[n]= X k=; W[k]s[n ; k] (2) where W[n] is the sequence of N N impulse response matrices corresponding to the separating system. We will denote G[n] = W[n] H[n] the matrix impulse response of the mixing and separation system. The aim in BSS is to nd or estimate W[n] suchthatthe separating system retrieves the original sources with some specic indeterminacies. The signal model is considerably simplied when instantaneous mixtures are considered. In this case H[n] = H[n] and W[n] = W[n], and equations () and (2) reduce to x = Hs (3) y = Wx (4) (Indices with respect to n are omitted in the instantaneous case to simplify notation). This paper is structured as follows. Section 2 presents the II method for instantaneous mixtures and a new

ON SOME EXTENSIONS OF THE NATURAL GRADIENT ALGORITHM. Brain Science Institute, RIKEN, Wako-shi, Saitama , Japan

ON SOME EXTENSIONS OF THE NATURAL GRADIENT ALGORITHM. Brain Science Institute, RIKEN, Wako-shi, Saitama , Japan ON SOME EXTENSIONS OF THE NATURAL GRADIENT ALGORITHM Pando Georgiev a, Andrzej Cichocki b and Shun-ichi Amari c Brain Science Institute, RIKEN, Wako-shi, Saitama 351-01, Japan a On leave from the Sofia

More information

MULTICHANNEL BLIND SEPARATION AND. Scott C. Douglas 1, Andrzej Cichocki 2, and Shun-ichi Amari 2

MULTICHANNEL BLIND SEPARATION AND. Scott C. Douglas 1, Andrzej Cichocki 2, and Shun-ichi Amari 2 MULTICHANNEL BLIND SEPARATION AND DECONVOLUTION OF SOURCES WITH ARBITRARY DISTRIBUTIONS Scott C. Douglas 1, Andrzej Cichoci, and Shun-ichi Amari 1 Department of Electrical Engineering, University of Utah

More information

1 Introduction Independent component analysis (ICA) [10] is a statistical technique whose main applications are blind source separation, blind deconvo

1 Introduction Independent component analysis (ICA) [10] is a statistical technique whose main applications are blind source separation, blind deconvo The Fixed-Point Algorithm and Maximum Likelihood Estimation for Independent Component Analysis Aapo Hyvarinen Helsinki University of Technology Laboratory of Computer and Information Science P.O.Box 5400,

More information

Massoud BABAIE-ZADEH. Blind Source Separation (BSS) and Independent Componen Analysis (ICA) p.1/39

Massoud BABAIE-ZADEH. Blind Source Separation (BSS) and Independent Componen Analysis (ICA) p.1/39 Blind Source Separation (BSS) and Independent Componen Analysis (ICA) Massoud BABAIE-ZADEH Blind Source Separation (BSS) and Independent Componen Analysis (ICA) p.1/39 Outline Part I Part II Introduction

More information

Natural Gradient Learning for Over- and Under-Complete Bases in ICA

Natural Gradient Learning for Over- and Under-Complete Bases in ICA NOTE Communicated by Jean-François Cardoso Natural Gradient Learning for Over- and Under-Complete Bases in ICA Shun-ichi Amari RIKEN Brain Science Institute, Wako-shi, Hirosawa, Saitama 351-01, Japan Independent

More information

Independent Component Analysis. Contents

Independent Component Analysis. Contents Contents Preface xvii 1 Introduction 1 1.1 Linear representation of multivariate data 1 1.1.1 The general statistical setting 1 1.1.2 Dimension reduction methods 2 1.1.3 Independence as a guiding principle

More information

To appear in Proceedings of the ICA'99, Aussois, France, A 2 R mn is an unknown mixture matrix of full rank, v(t) is the vector of noises. The

To appear in Proceedings of the ICA'99, Aussois, France, A 2 R mn is an unknown mixture matrix of full rank, v(t) is the vector of noises. The To appear in Proceedings of the ICA'99, Aussois, France, 1999 1 NATURAL GRADIENT APPROACH TO BLIND SEPARATION OF OVER- AND UNDER-COMPLETE MIXTURES L.-Q. Zhang, S. Amari and A. Cichocki Brain-style Information

More information

x 1 (t) Spectrogram t s

x 1 (t) Spectrogram t s A METHOD OF ICA IN TIME-FREQUENCY DOMAIN Shiro Ikeda PRESTO, JST Hirosawa 2-, Wako, 35-98, Japan Shiro.Ikeda@brain.riken.go.jp Noboru Murata RIKEN BSI Hirosawa 2-, Wako, 35-98, Japan Noboru.Murata@brain.riken.go.jp

More information

1 Introduction Blind source separation (BSS) is a fundamental problem which is encountered in a variety of signal processing problems where multiple s

1 Introduction Blind source separation (BSS) is a fundamental problem which is encountered in a variety of signal processing problems where multiple s Blind Separation of Nonstationary Sources in Noisy Mixtures Seungjin CHOI x1 and Andrzej CICHOCKI y x Department of Electrical Engineering Chungbuk National University 48 Kaeshin-dong, Cheongju Chungbuk

More information

Analytical solution of the blind source separation problem using derivatives

Analytical solution of the blind source separation problem using derivatives Analytical solution of the blind source separation problem using derivatives Sebastien Lagrange 1,2, Luc Jaulin 2, Vincent Vigneron 1, and Christian Jutten 1 1 Laboratoire Images et Signaux, Institut National

More information

where A 2 IR m n is the mixing matrix, s(t) is the n-dimensional source vector (n» m), and v(t) is additive white noise that is statistically independ

where A 2 IR m n is the mixing matrix, s(t) is the n-dimensional source vector (n» m), and v(t) is additive white noise that is statistically independ BLIND SEPARATION OF NONSTATIONARY AND TEMPORALLY CORRELATED SOURCES FROM NOISY MIXTURES Seungjin CHOI x and Andrzej CICHOCKI y x Department of Electrical Engineering Chungbuk National University, KOREA

More information

MINIMIZATION-PROJECTION (MP) APPROACH FOR BLIND SOURCE SEPARATION IN DIFFERENT MIXING MODELS

MINIMIZATION-PROJECTION (MP) APPROACH FOR BLIND SOURCE SEPARATION IN DIFFERENT MIXING MODELS MINIMIZATION-PROJECTION (MP) APPROACH FOR BLIND SOURCE SEPARATION IN DIFFERENT MIXING MODELS Massoud Babaie-Zadeh ;2, Christian Jutten, Kambiz Nayebi 2 Institut National Polytechnique de Grenoble (INPG),

More information

On Information Maximization and Blind Signal Deconvolution

On Information Maximization and Blind Signal Deconvolution On Information Maximization and Blind Signal Deconvolution A Röbel Technical University of Berlin, Institute of Communication Sciences email: roebel@kgwtu-berlinde Abstract: In the following paper we investigate

More information

Blind separation of instantaneous mixtures of dependent sources

Blind separation of instantaneous mixtures of dependent sources Blind separation of instantaneous mixtures of dependent sources Marc Castella and Pierre Comon GET/INT, UMR-CNRS 7, 9 rue Charles Fourier, 9 Évry Cedex, France marc.castella@int-evry.fr, CNRS, I3S, UMR

More information

Blind Source Separation with a Time-Varying Mixing Matrix

Blind Source Separation with a Time-Varying Mixing Matrix Blind Source Separation with a Time-Varying Mixing Matrix Marcus R DeYoung and Brian L Evans Department of Electrical and Computer Engineering The University of Texas at Austin 1 University Station, Austin,

More information

Fundamentals of Principal Component Analysis (PCA), Independent Component Analysis (ICA), and Independent Vector Analysis (IVA)

Fundamentals of Principal Component Analysis (PCA), Independent Component Analysis (ICA), and Independent Vector Analysis (IVA) Fundamentals of Principal Component Analysis (PCA),, and Independent Vector Analysis (IVA) Dr Mohsen Naqvi Lecturer in Signal and Information Processing, School of Electrical and Electronic Engineering,

More information

a 22(t) a nn(t) Correlation τ=0 τ=1 τ=2 (t) x 1 (t) x Time(sec)

a 22(t) a nn(t) Correlation τ=0 τ=1 τ=2 (t) x 1 (t) x Time(sec) A METHOD OF BLIND SEPARATION ON TEMPORAL STRUCTURE OF SIGNALS Shiro Ikeda and Noboru Murata Email:fShiro.Ikeda,Noboru.Muratag@brain.riken.go.jp RIKEN Brain Science Institute Hirosawa -, Wako, Saitama 3-98,

More information

1 Introduction The Separation of Independent Sources (SIS) assumes that some unknown but independent temporal signals propagate through a mixing and/o

1 Introduction The Separation of Independent Sources (SIS) assumes that some unknown but independent temporal signals propagate through a mixing and/o Appeared in IEEE Trans. on Circuits and Systems, vol. 42, no. 11, pp. 748-751, November 95 c Implementation and Test Results of a Chip for The Separation of Mixed Signals Ammar B. A. Gharbi and Fathi M.

More information

Semi-Blind approaches to source separation: introduction to the special session

Semi-Blind approaches to source separation: introduction to the special session Semi-Blind approaches to source separation: introduction to the special session Massoud BABAIE-ZADEH 1 Christian JUTTEN 2 1- Sharif University of Technology, Tehran, IRAN 2- Laboratory of Images and Signals

More information

ICA [6] ICA) [7, 8] ICA ICA ICA [9, 10] J-F. Cardoso. [13] Matlab ICA. Comon[3], Amari & Cardoso[4] ICA ICA

ICA [6] ICA) [7, 8] ICA ICA ICA [9, 10] J-F. Cardoso. [13] Matlab ICA. Comon[3], Amari & Cardoso[4] ICA ICA 16 1 (Independent Component Analysis: ICA) 198 9 ICA ICA ICA 1 ICA 198 Jutten Herault Comon[3], Amari & Cardoso[4] ICA Comon (PCA) projection persuit projection persuit ICA ICA ICA 1 [1] [] ICA ICA EEG

More information

ORIENTED PCA AND BLIND SIGNAL SEPARATION

ORIENTED PCA AND BLIND SIGNAL SEPARATION ORIENTED PCA AND BLIND SIGNAL SEPARATION K. I. Diamantaras Department of Informatics TEI of Thessaloniki Sindos 54101, Greece kdiamant@it.teithe.gr Th. Papadimitriou Department of Int. Economic Relat.

More information

An Iterative Blind Source Separation Method for Convolutive Mixtures of Images

An Iterative Blind Source Separation Method for Convolutive Mixtures of Images An Iterative Blind Source Separation Method for Convolutive Mixtures of Images Marc Castella and Jean-Christophe Pesquet Université de Marne-la-Vallée / UMR-CNRS 8049 5 bd Descartes, Champs-sur-Marne 77454

More information

A METHOD OF ICA IN TIME-FREQUENCY DOMAIN

A METHOD OF ICA IN TIME-FREQUENCY DOMAIN A METHOD OF ICA IN TIME-FREQUENCY DOMAIN Shiro Ikeda PRESTO, JST Hirosawa 2-, Wako, 35-98, Japan Shiro.Ikeda@brain.riken.go.jp Noboru Murata RIKEN BSI Hirosawa 2-, Wako, 35-98, Japan Noboru.Murata@brain.riken.go.jp

More information

ON-LINE BLIND SEPARATION OF NON-STATIONARY SIGNALS

ON-LINE BLIND SEPARATION OF NON-STATIONARY SIGNALS Yugoslav Journal of Operations Research 5 (25), Number, 79-95 ON-LINE BLIND SEPARATION OF NON-STATIONARY SIGNALS Slavica TODOROVIĆ-ZARKULA EI Professional Electronics, Niš, bssmtod@eunet.yu Branimir TODOROVIĆ,

More information

PROPERTIES OF THE EMPIRICAL CHARACTERISTIC FUNCTION AND ITS APPLICATION TO TESTING FOR INDEPENDENCE. Noboru Murata

PROPERTIES OF THE EMPIRICAL CHARACTERISTIC FUNCTION AND ITS APPLICATION TO TESTING FOR INDEPENDENCE. Noboru Murata ' / PROPERTIES OF THE EMPIRICAL CHARACTERISTIC FUNCTION AND ITS APPLICATION TO TESTING FOR INDEPENDENCE Noboru Murata Waseda University Department of Electrical Electronics and Computer Engineering 3--

More information

SPARSE REPRESENTATION AND BLIND DECONVOLUTION OF DYNAMICAL SYSTEMS. Liqing Zhang and Andrzej Cichocki

SPARSE REPRESENTATION AND BLIND DECONVOLUTION OF DYNAMICAL SYSTEMS. Liqing Zhang and Andrzej Cichocki SPARSE REPRESENTATON AND BLND DECONVOLUTON OF DYNAMCAL SYSTEMS Liqing Zhang and Andrzej Cichocki Lab for Advanced Brain Signal Processing RKEN Brain Science nstitute Wako shi, Saitama, 351-198, apan zha,cia

More information

Undercomplete Independent Component. Analysis for Signal Separation and. Dimension Reduction. Category: Algorithms and Architectures.

Undercomplete Independent Component. Analysis for Signal Separation and. Dimension Reduction. Category: Algorithms and Architectures. Undercomplete Independent Component Analysis for Signal Separation and Dimension Reduction John Porrill and James V Stone Psychology Department, Sheeld University, Sheeld, S10 2UR, England. Tel: 0114 222

More information

POLYNOMIAL SINGULAR VALUES FOR NUMBER OF WIDEBAND SOURCES ESTIMATION AND PRINCIPAL COMPONENT ANALYSIS

POLYNOMIAL SINGULAR VALUES FOR NUMBER OF WIDEBAND SOURCES ESTIMATION AND PRINCIPAL COMPONENT ANALYSIS POLYNOMIAL SINGULAR VALUES FOR NUMBER OF WIDEBAND SOURCES ESTIMATION AND PRINCIPAL COMPONENT ANALYSIS Russell H. Lambert RF and Advanced Mixed Signal Unit Broadcom Pasadena, CA USA russ@broadcom.com Marcel

More information

A Canonical Genetic Algorithm for Blind Inversion of Linear Channels

A Canonical Genetic Algorithm for Blind Inversion of Linear Channels A Canonical Genetic Algorithm for Blind Inversion of Linear Channels Fernando Rojas, Jordi Solé-Casals, Enric Monte-Moreno 3, Carlos G. Puntonet and Alberto Prieto Computer Architecture and Technology

More information

Independent Component Analysis of Incomplete Data

Independent Component Analysis of Incomplete Data Independent Component Analysis of Incomplete Data Max Welling Markus Weber California Institute of Technology 136-93 Pasadena, CA 91125 fwelling,rmwg@vision.caltech.edu Keywords: EM, Missing Data, ICA

More information

Recursive Generalized Eigendecomposition for Independent Component Analysis

Recursive Generalized Eigendecomposition for Independent Component Analysis Recursive Generalized Eigendecomposition for Independent Component Analysis Umut Ozertem 1, Deniz Erdogmus 1,, ian Lan 1 CSEE Department, OGI, Oregon Health & Science University, Portland, OR, USA. {ozertemu,deniz}@csee.ogi.edu

More information

BLIND SEPARATION OF INSTANTANEOUS MIXTURES OF NON STATIONARY SOURCES

BLIND SEPARATION OF INSTANTANEOUS MIXTURES OF NON STATIONARY SOURCES BLIND SEPARATION OF INSTANTANEOUS MIXTURES OF NON STATIONARY SOURCES Dinh-Tuan Pham Laboratoire de Modélisation et Calcul URA 397, CNRS/UJF/INPG BP 53X, 38041 Grenoble cédex, France Dinh-Tuan.Pham@imag.fr

More information

One-unit Learning Rules for Independent Component Analysis

One-unit Learning Rules for Independent Component Analysis One-unit Learning Rules for Independent Component Analysis Aapo Hyvarinen and Erkki Oja Helsinki University of Technology Laboratory of Computer and Information Science Rakentajanaukio 2 C, FIN-02150 Espoo,

More information

Non-Euclidean Independent Component Analysis and Oja's Learning

Non-Euclidean Independent Component Analysis and Oja's Learning Non-Euclidean Independent Component Analysis and Oja's Learning M. Lange 1, M. Biehl 2, and T. Villmann 1 1- University of Appl. Sciences Mittweida - Dept. of Mathematics Mittweida, Saxonia - Germany 2-

More information

Independent Component Analysis (ICA) Bhaskar D Rao University of California, San Diego

Independent Component Analysis (ICA) Bhaskar D Rao University of California, San Diego Independent Component Analysis (ICA) Bhaskar D Rao University of California, San Diego Email: brao@ucsdedu References 1 Hyvarinen, A, Karhunen, J, & Oja, E (2004) Independent component analysis (Vol 46)

More information

Blind Source Separation in Nonlinear Mixture for Colored Sources Using Signal Derivatives

Blind Source Separation in Nonlinear Mixture for Colored Sources Using Signal Derivatives Blind Source Separation in Nonlinear Mixture for Colored Sources Using Signal Derivatives Bahram Ehsandoust, Masoud Babaie-Zadeh, Christian Jutten To cite this version: Bahram Ehsandoust, Masoud Babaie-Zadeh,

More information

CCA BASED ALGORITHMS FOR BLIND EQUALIZATION OF FIR MIMO SYSTEMS

CCA BASED ALGORITHMS FOR BLIND EQUALIZATION OF FIR MIMO SYSTEMS CCA BASED ALGORITHMS FOR BLID EQUALIZATIO OF FIR MIMO SYSTEMS Javier Vía and Ignacio Santamaría Dept of Communications Engineering University of Cantabria 395 Santander, Cantabria, Spain E-mail: {jvia,nacho}@gtasdicomunicanes

More information

Linear Regression and Its Applications

Linear Regression and Its Applications Linear Regression and Its Applications Predrag Radivojac October 13, 2014 Given a data set D = {(x i, y i )} n the objective is to learn the relationship between features and the target. We usually start

More information

Blind Machine Separation Te-Won Lee

Blind Machine Separation Te-Won Lee Blind Machine Separation Te-Won Lee University of California, San Diego Institute for Neural Computation Blind Machine Separation Problem we want to solve: Single microphone blind source separation & deconvolution

More information

An Improved Cumulant Based Method for Independent Component Analysis

An Improved Cumulant Based Method for Independent Component Analysis An Improved Cumulant Based Method for Independent Component Analysis Tobias Blaschke and Laurenz Wiskott Institute for Theoretical Biology Humboldt University Berlin Invalidenstraße 43 D - 0 5 Berlin Germany

More information

BLIND SEPARATION USING ABSOLUTE MOMENTS BASED ADAPTIVE ESTIMATING FUNCTION. Juha Karvanen and Visa Koivunen

BLIND SEPARATION USING ABSOLUTE MOMENTS BASED ADAPTIVE ESTIMATING FUNCTION. Juha Karvanen and Visa Koivunen BLIND SEPARATION USING ABSOLUTE MOMENTS BASED ADAPTIVE ESTIMATING UNCTION Juha Karvanen and Visa Koivunen Signal Processing Laboratory Helsinki University of Technology P.O. Box 3, IN-215 HUT, inland Tel.

More information

Wavelet de-noising for blind source separation in noisy mixtures.

Wavelet de-noising for blind source separation in noisy mixtures. Wavelet for blind source separation in noisy mixtures. Bertrand Rivet 1, Vincent Vigneron 1, Anisoara Paraschiv-Ionescu 2 and Christian Jutten 1 1 Institut National Polytechnique de Grenoble. Laboratoire

More information

CRITERIA FOR THE SIMULTANEOUS BLIND EXTRACTION OF ARBITRARY GROUPS OF SOURCES

CRITERIA FOR THE SIMULTANEOUS BLIND EXTRACTION OF ARBITRARY GROUPS OF SOURCES CRITRI OR TH SIMULTNOUS BLIN XTRCTION O RBITRRY GROUPS O SOURCS Sergio Cruces Signal Processing Group Camino escubrimientos, 41092-Seville, Spain sergio@cicaes ndrzej Cichocki, Shun-ichi mari Brain Science

More information

TRINICON: A Versatile Framework for Multichannel Blind Signal Processing

TRINICON: A Versatile Framework for Multichannel Blind Signal Processing TRINICON: A Versatile Framework for Multichannel Blind Signal Processing Herbert Buchner, Robert Aichner, Walter Kellermann {buchner,aichner,wk}@lnt.de Telecommunications Laboratory University of Erlangen-Nuremberg

More information

Estimation of linear non-gaussian acyclic models for latent factors

Estimation of linear non-gaussian acyclic models for latent factors Estimation of linear non-gaussian acyclic models for latent factors Shohei Shimizu a Patrik O. Hoyer b Aapo Hyvärinen b,c a The Institute of Scientific and Industrial Research, Osaka University Mihogaoka

More information

Blind channel deconvolution of real world signals using source separation techniques

Blind channel deconvolution of real world signals using source separation techniques Blind channel deconvolution of real world signals using source separation techniques Jordi Solé-Casals 1, Enric Monte-Moreno 2 1 Signal Processing Group, University of Vic, Sagrada Família 7, 08500, Vic

More information

FREQUENCY-DOMAIN IMPLEMENTATION OF BLOCK ADAPTIVE FILTERS FOR ICA-BASED MULTICHANNEL BLIND DECONVOLUTION

FREQUENCY-DOMAIN IMPLEMENTATION OF BLOCK ADAPTIVE FILTERS FOR ICA-BASED MULTICHANNEL BLIND DECONVOLUTION FREQUENCY-DOMAIN IMPLEMENTATION OF BLOCK ADAPTIVE FILTERS FOR ICA-BASED MULTICHANNEL BLIND DECONVOLUTION Kyungmin Na, Sang-Chul Kang, Kyung Jin Lee, and Soo-Ik Chae School of Electrical Engineering Seoul

More information

Simultaneous Diagonalization in the Frequency Domain (SDIF) for Source Separation

Simultaneous Diagonalization in the Frequency Domain (SDIF) for Source Separation Simultaneous Diagonalization in the Frequency Domain (SDIF) for Source Separation Hsiao-Chun Wu and Jose C. Principe Computational Neuro-Engineering Laboratory Department of Electrical and Computer Engineering

More information

Blind Source Separation for Changing Source Number: A Neural Network Approach with a Variable Structure 1

Blind Source Separation for Changing Source Number: A Neural Network Approach with a Variable Structure 1 Blind Source Separation for Changing Source Number: A Neural Network Approach with a Variable Structure 1 Shun-Tian Lou, Member, IEEE and Xian-Da Zhang, Senior Member, IEEE Key Lab for Radar Signal Processing,

More information

ICA ALGORITHM BASED ON SUBSPACE APPROACH. Ali MANSOUR, Member of IEEE and N. Ohnishi,Member of IEEE. Bio-Mimetic Control Research Center (RIKEN),

ICA ALGORITHM BASED ON SUBSPACE APPROACH. Ali MANSOUR, Member of IEEE and N. Ohnishi,Member of IEEE. Bio-Mimetic Control Research Center (RIKEN), ICA ALGORITHM BASED ON SUBSPACE APPROACH Ali MANSOUR, Member of IEEE and N Ohnishi,Member of IEEE Bio-Mimetic Control Research Center (RIKEN), 7-, Anagahora, Shimoshidami, Moriyama-ku, Nagoya (JAPAN) mail:mansournagoyarikengojp,ohnishisheratonohnishinuienagoya-uacjp

More information

Independent Component Analysis on the Basis of Helmholtz Machine

Independent Component Analysis on the Basis of Helmholtz Machine Independent Component Analysis on the Basis of Helmholtz Machine Masashi OHATA *1 ohatama@bmc.riken.go.jp Toshiharu MUKAI *1 tosh@bmc.riken.go.jp Kiyotoshi MATSUOKA *2 matsuoka@brain.kyutech.ac.jp *1 Biologically

More information

g(.) 1/ N 1/ N Decision Decision Device u u u u CP

g(.) 1/ N 1/ N Decision Decision Device u u u u CP Distributed Weak Signal Detection and Asymptotic Relative Eciency in Dependent Noise Hakan Delic Signal and Image Processing Laboratory (BUSI) Department of Electrical and Electronics Engineering Bogazici

More information

Nonholonomic Orthogonal Learning Algorithms for Blind Source Separation

Nonholonomic Orthogonal Learning Algorithms for Blind Source Separation LETTER Communicated by Klaus Obermayer Nonholonomic Orthogonal Learning Algorithms for Blind Source Separation Shun-ichi Amari Tian-Ping Chen Andrzej Cichocki RIKEN Brain Science Institute, Brain-Style

More information

Non-Negative Matrix Factorization with Quasi-Newton Optimization

Non-Negative Matrix Factorization with Quasi-Newton Optimization Non-Negative Matrix Factorization with Quasi-Newton Optimization Rafal ZDUNEK, Andrzej CICHOCKI Laboratory for Advanced Brain Signal Processing BSI, RIKEN, Wako-shi, JAPAN Abstract. Non-negative matrix

More information

Principal Component Analysis

Principal Component Analysis Principal Component Analysis Introduction Consider a zero mean random vector R n with autocorrelation matri R = E( T ). R has eigenvectors q(1),,q(n) and associated eigenvalues λ(1) λ(n). Let Q = [ q(1)

More information

Gatsby Theoretical Neuroscience Lectures: Non-Gaussian statistics and natural images Parts I-II

Gatsby Theoretical Neuroscience Lectures: Non-Gaussian statistics and natural images Parts I-II Gatsby Theoretical Neuroscience Lectures: Non-Gaussian statistics and natural images Parts I-II Gatsby Unit University College London 27 Feb 2017 Outline Part I: Theory of ICA Definition and difference

More information

1 Introduction Consider the following: given a cost function J (w) for the parameter vector w = [w1 w2 w n ] T, maximize J (w) (1) such that jjwjj = C

1 Introduction Consider the following: given a cost function J (w) for the parameter vector w = [w1 w2 w n ] T, maximize J (w) (1) such that jjwjj = C On Gradient Adaptation With Unit-Norm Constraints Scott C. Douglas 1, Shun-ichi Amari 2, and S.-Y. Kung 3 1 Department of Electrical Engineering, Southern Methodist University Dallas, Texas 75275 USA 2

More information

Blind Deconvolution by Modified Bussgang Algorithm

Blind Deconvolution by Modified Bussgang Algorithm Reprinted from ISCAS'99, International Symposium on Circuits and Systems, Orlando, Florida, May 3-June 2, 1999 Blind Deconvolution by Modified Bussgang Algorithm Simone Fiori, Aurelio Uncini, Francesco

More information

Complete Blind Subspace Deconvolution

Complete Blind Subspace Deconvolution Complete Blind Subspace Deconvolution Zoltán Szabó Department of Information Systems, Eötvös Loránd University, Pázmány P. sétány 1/C, Budapest H-1117, Hungary szzoli@cs.elte.hu http://nipg.inf.elte.hu

More information

A Constrained EM Algorithm for Independent Component Analysis

A Constrained EM Algorithm for Independent Component Analysis LETTER Communicated by Hagai Attias A Constrained EM Algorithm for Independent Component Analysis Max Welling Markus Weber California Institute of Technology, Pasadena, CA 91125, U.S.A. We introduce a

More information

HST.582J / 6.555J / J Biomedical Signal and Image Processing Spring 2007

HST.582J / 6.555J / J Biomedical Signal and Image Processing Spring 2007 MIT OpenCourseWare http://ocw.mit.edu HST.582J / 6.555J / 16.456J Biomedical Signal and Image Processing Spring 2007 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.

More information

BLIND SOURCE SEPARATION IN POST NONLINEAR MIXTURES

BLIND SOURCE SEPARATION IN POST NONLINEAR MIXTURES BLIND SURCE SEPARATIN IN PST NNLINEAR MIXTURES Sophie Achard, Dinh Tuan Pham Univ of Grenoble Laboratory of Modeling Computation, IMAG, CNRS BP 53X, 384 Grenoble Cedex, France SophieAchard@imagfr, DinhTuanPham@imagfr

More information

BLIND SEPARATION OF SOURCES: A NON-LINEAR NEURAL ALGORITHM

BLIND SEPARATION OF SOURCES: A NON-LINEAR NEURAL ALGORITHM BLIND SEPARATION OF SOURCES: A NON-LINEAR NEURAL ALGORITHM Gilles BUREL Thomson CSF, Laboratoires Electroniques de Rennes, Avenue de Belle Fontaine, F-35510 CESSON-SEVIGNE, France My new address is: Prof.

More information

A Generalization of Blind Source Separation Algorithms for Convolutive Mixtures Based on Second-Order Statistics

A Generalization of Blind Source Separation Algorithms for Convolutive Mixtures Based on Second-Order Statistics 120 IEEE TRANSACTIONS ON SPEECH AND AUDIO PROCESSING, VOL 13, NO 1, JANUARY 2005 A Generalization of Blind Source Separation Algorithms for Convolutive Mixtures Based on Second-Order Statistics Herbert

More information

ACENTRAL problem in neural-network research, as well

ACENTRAL problem in neural-network research, as well 626 IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 10, NO. 3, MAY 1999 Fast and Robust Fixed-Point Algorithms for Independent Component Analysis Aapo Hyvärinen Abstract Independent component analysis (ICA)

More information

Robust extraction of specific signals with temporal structure

Robust extraction of specific signals with temporal structure Robust extraction of specific signals with temporal structure Zhi-Lin Zhang, Zhang Yi Computational Intelligence Laboratory, School of Computer Science and Engineering, University of Electronic Science

More information

Examples. 2-input, 1-output discrete-time systems: 1-input, 1-output discrete-time systems:

Examples. 2-input, 1-output discrete-time systems: 1-input, 1-output discrete-time systems: Discrete-Time s - I Time-Domain Representation CHAPTER 4 These lecture slides are based on "Digital Signal Processing: A Computer-Based Approach, 4th ed." textbook by S.K. Mitra and its instructor materials.

More information

Blind Extraction of Singularly Mixed Source Signals

Blind Extraction of Singularly Mixed Source Signals IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL 11, NO 6, NOVEMBER 2000 1413 Blind Extraction of Singularly Mixed Source Signals Yuanqing Li, Jun Wang, Senior Member, IEEE, and Jacek M Zurada, Fellow, IEEE Abstract

More information

Neural networks for blind separation with unknown number of sources

Neural networks for blind separation with unknown number of sources Neurocomputing 24 (1999) 55 93 Neural networks for blind separation with unknown number of sources Andrzej Cichocki*, Juha Karhunen, Wlodzimierz Kasprzak, Ricardo Vigário Laboratory for Open Information

More information

Blind signal processing algorithms

Blind signal processing algorithms 12th Int. Workshop on Systems, Signals & Image Processing, 22-24 September 2005, Chalkida, Greece 105 Blind signal processing algorithms Athanasios Margaris and Efthimios Kotsialos Department of Applied

More information

Blind Deconvolution via Maximum Kurtosis Adaptive Filtering

Blind Deconvolution via Maximum Kurtosis Adaptive Filtering Blind Deconvolution via Maximum Kurtosis Adaptive Filtering Deborah Pereg Doron Benzvi The Jerusalem College of Engineering Jerusalem, Israel doronb@jce.ac.il, deborahpe@post.jce.ac.il ABSTRACT In this

More information

ADAPTIVE FILTER THEORY

ADAPTIVE FILTER THEORY ADAPTIVE FILTER THEORY Fourth Edition Simon Haykin Communications Research Laboratory McMaster University Hamilton, Ontario, Canada Front ice Hall PRENTICE HALL Upper Saddle River, New Jersey 07458 Preface

More information

Blind Source Separation Using Second-Order Cyclostationary Statistics

Blind Source Separation Using Second-Order Cyclostationary Statistics 694 IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 49, NO. 4, APRIL 2001 Blind Source Separation Using Second-Order Cyclostationary Statistics Karim Abed-Meraim, Yong Xiang, Jonathan H. Manton, Yingbo Hua,

More information

Blind Signal Separation: Statistical Principles

Blind Signal Separation: Statistical Principles Blind Signal Separation: Statistical Principles JEAN-FRANÇOIS CARDOSO, MEMBER, IEEE Invited Paper Blind signal separation (BSS) and independent component analysis (ICA) are emerging techniques of array

More information

CONTROL SYSTEMS ANALYSIS VIA BLIND SOURCE DECONVOLUTION. Kenji Sugimoto and Yoshito Kikkawa

CONTROL SYSTEMS ANALYSIS VIA BLIND SOURCE DECONVOLUTION. Kenji Sugimoto and Yoshito Kikkawa CONTROL SYSTEMS ANALYSIS VIA LIND SOURCE DECONVOLUTION Kenji Sugimoto and Yoshito Kikkawa Nara Institute of Science and Technology Graduate School of Information Science 896-5 Takayama-cho, Ikoma-city,

More information

CIFAR Lectures: Non-Gaussian statistics and natural images

CIFAR Lectures: Non-Gaussian statistics and natural images CIFAR Lectures: Non-Gaussian statistics and natural images Dept of Computer Science University of Helsinki, Finland Outline Part I: Theory of ICA Definition and difference to PCA Importance of non-gaussianity

More information

Independent Component Analysis

Independent Component Analysis A Short Introduction to Independent Component Analysis with Some Recent Advances Aapo Hyvärinen Dept of Computer Science Dept of Mathematics and Statistics University of Helsinki Problem of blind source

More information

Dimensionality Reduction. CS57300 Data Mining Fall Instructor: Bruno Ribeiro

Dimensionality Reduction. CS57300 Data Mining Fall Instructor: Bruno Ribeiro Dimensionality Reduction CS57300 Data Mining Fall 2016 Instructor: Bruno Ribeiro Goal } Visualize high dimensional data (and understand its Geometry) } Project the data into lower dimensional spaces }

More information

FUNDAMENTAL LIMITATION OF FREQUENCY DOMAIN BLIND SOURCE SEPARATION FOR CONVOLVED MIXTURE OF SPEECH

FUNDAMENTAL LIMITATION OF FREQUENCY DOMAIN BLIND SOURCE SEPARATION FOR CONVOLVED MIXTURE OF SPEECH FUNDAMENTAL LIMITATION OF FREQUENCY DOMAIN BLIND SOURCE SEPARATION FOR CONVOLVED MIXTURE OF SPEECH Shoko Araki Shoji Makino Ryo Mukai Tsuyoki Nishikawa Hiroshi Saruwatari NTT Communication Science Laboratories

More information

Remaining energy on log scale Number of linear PCA components

Remaining energy on log scale Number of linear PCA components NONLINEAR INDEPENDENT COMPONENT ANALYSIS USING ENSEMBLE LEARNING: EXPERIMENTS AND DISCUSSION Harri Lappalainen, Xavier Giannakopoulos, Antti Honkela, and Juha Karhunen Helsinki University of Technology,

More information

ARTIFICIAL INTELLIGENCE LABORATORY. and CENTER FOR BIOLOGICAL INFORMATION PROCESSING. A.I. Memo No August Federico Girosi.

ARTIFICIAL INTELLIGENCE LABORATORY. and CENTER FOR BIOLOGICAL INFORMATION PROCESSING. A.I. Memo No August Federico Girosi. MASSACHUSETTS INSTITUTE OF TECHNOLOGY ARTIFICIAL INTELLIGENCE LABORATORY and CENTER FOR BIOLOGICAL INFORMATION PROCESSING WHITAKER COLLEGE A.I. Memo No. 1287 August 1991 C.B.I.P. Paper No. 66 Models of

More information

CONVOLUTIVE NON-NEGATIVE MATRIX FACTORISATION WITH SPARSENESS CONSTRAINT

CONVOLUTIVE NON-NEGATIVE MATRIX FACTORISATION WITH SPARSENESS CONSTRAINT CONOLUTIE NON-NEGATIE MATRIX FACTORISATION WITH SPARSENESS CONSTRAINT Paul D. O Grady Barak A. Pearlmutter Hamilton Institute National University of Ireland, Maynooth Co. Kildare, Ireland. ABSTRACT Discovering

More information

Convolutive Blind Source Separation based on Multiple Decorrelation. Lucas Parra, Clay Spence, Bert De Vries Sarno Corporation, CN-5300, Princeton, NJ

Convolutive Blind Source Separation based on Multiple Decorrelation. Lucas Parra, Clay Spence, Bert De Vries Sarno Corporation, CN-5300, Princeton, NJ Convolutive Blind Source Separation based on Multiple Decorrelation. Lucas Parra, Clay Spence, Bert De Vries Sarno Corporation, CN-5300, Princeton, NJ 08543 lparra j cspence j bdevries @ sarno.com Abstract

More information

Independent Component Analysis and Its Applications. By Qing Xue, 10/15/2004

Independent Component Analysis and Its Applications. By Qing Xue, 10/15/2004 Independent Component Analysis and Its Applications By Qing Xue, 10/15/2004 Outline Motivation of ICA Applications of ICA Principles of ICA estimation Algorithms for ICA Extensions of basic ICA framework

More information

On the INFOMAX algorithm for blind signal separation

On the INFOMAX algorithm for blind signal separation University of Wollongong Research Online Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences 2000 On the INFOMAX algorithm for blind signal separation Jiangtao Xi

More information

Tutorial on Blind Source Separation and Independent Component Analysis

Tutorial on Blind Source Separation and Independent Component Analysis Tutorial on Blind Source Separation and Independent Component Analysis Lucas Parra Adaptive Image & Signal Processing Group Sarnoff Corporation February 09, 2002 Linear Mixtures... problem statement...

More information

Independent Component Analysis

Independent Component Analysis A Short Introduction to Independent Component Analysis Aapo Hyvärinen Helsinki Institute for Information Technology and Depts of Computer Science and Psychology University of Helsinki Problem of blind

More information

BLIND SEPARATION OF INSTANTANEOUS MIXTURES OF NON STATIONARY SOURCES

BLIND SEPARATION OF INSTANTANEOUS MIXTURES OF NON STATIONARY SOURCES BLIND SEPARATION OF INSTANTANEOUS MIXTURES OF NON STATIONARY SOURCES Dinh-Tuan Pham Laboratoire de Modélisation et Calcul URA 397, CNRS/UJF/INPG BP 53X, 38041 Grenoble cédex, France Dinh-Tuan.Pham@imag.fr

More information

Comparative Analysis of ICA Based Features

Comparative Analysis of ICA Based Features International Journal of Emerging Engineering Research and Technology Volume 2, Issue 7, October 2014, PP 267-273 ISSN 2349-4395 (Print) & ISSN 2349-4409 (Online) Comparative Analysis of ICA Based Features

More information

FASTGEO - A HISTOGRAM BASED APPROACH TO LINEAR GEOMETRIC ICA. Andreas Jung, Fabian J. Theis, Carlos G. Puntonet, Elmar W. Lang

FASTGEO - A HISTOGRAM BASED APPROACH TO LINEAR GEOMETRIC ICA. Andreas Jung, Fabian J. Theis, Carlos G. Puntonet, Elmar W. Lang FASTGEO - A HISTOGRAM BASED APPROACH TO LINEAR GEOMETRIC ICA Andreas Jung, Fabian J Theis, Carlos G Puntonet, Elmar W Lang Institute for Theoretical Physics, University of Regensburg Institute of Biophysics,

More information

HST.582J/6.555J/16.456J

HST.582J/6.555J/16.456J Blind Source Separation: PCA & ICA HST.582J/6.555J/16.456J Gari D. Clifford gari [at] mit. edu http://www.mit.edu/~gari G. D. Clifford 2005-2009 What is BSS? Assume an observation (signal) is a linear

More information

A UNIFIED PRESENTATION OF BLIND SEPARATION METHODS FOR CONVOLUTIVE MIXTURES USING BLOCK-DIAGONALIZATION

A UNIFIED PRESENTATION OF BLIND SEPARATION METHODS FOR CONVOLUTIVE MIXTURES USING BLOCK-DIAGONALIZATION A UNIFIED PRESENTATION OF BLIND SEPARATION METHODS FOR CONVOLUTIVE MIXTURES USING BLOCK-DIAGONALIZATION Cédric Févotte and Christian Doncarli Institut de Recherche en Communications et Cybernétique de

More information

Distinguishing Causes from Effects using Nonlinear Acyclic Causal Models

Distinguishing Causes from Effects using Nonlinear Acyclic Causal Models JMLR Workshop and Conference Proceedings 6:17 164 NIPS 28 workshop on causality Distinguishing Causes from Effects using Nonlinear Acyclic Causal Models Kun Zhang Dept of Computer Science and HIIT University

More information

Chapter 6 Effective Blind Source Separation Based on the Adam Algorithm

Chapter 6 Effective Blind Source Separation Based on the Adam Algorithm Chapter 6 Effective Blind Source Separation Based on the Adam Algorithm Michele Scarpiniti, Simone Scardapane, Danilo Comminiello, Raffaele Parisi and Aurelio Uncini Abstract In this paper, we derive a

More information

Distinguishing Causes from Effects using Nonlinear Acyclic Causal Models

Distinguishing Causes from Effects using Nonlinear Acyclic Causal Models Distinguishing Causes from Effects using Nonlinear Acyclic Causal Models Kun Zhang Dept of Computer Science and HIIT University of Helsinki 14 Helsinki, Finland kun.zhang@cs.helsinki.fi Aapo Hyvärinen

More information

Lecture'12:' SSMs;'Independent'Component'Analysis;' Canonical'Correla;on'Analysis'

Lecture'12:' SSMs;'Independent'Component'Analysis;' Canonical'Correla;on'Analysis' Lecture'12:' SSMs;'Independent'Component'Analysis;' Canonical'Correla;on'Analysis' Lester'Mackey' May'7,'2014' ' Stats'306B:'Unsupervised'Learning' Beyond'linearity'in'state'space'modeling' Credit:'Alex'Simma'

More information

Speed and Accuracy Enhancement of Linear ICA Techniques Using Rational Nonlinear Functions

Speed and Accuracy Enhancement of Linear ICA Techniques Using Rational Nonlinear Functions Speed and Accuracy Enhancement of Linear ICA Techniques Using Rational Nonlinear Functions Petr Tichavský 1, Zbyněk Koldovský 1,2, and Erkki Oja 3 1 Institute of Information Theory and Automation, Pod

More information

Nonlinear Blind Source Separation Using a Radial Basis Function Network

Nonlinear Blind Source Separation Using a Radial Basis Function Network 124 IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 12, NO. 1, JANUARY 2001 Nonlinear Blind Source Separation Using a Radial Basis Function Network Ying Tan, Member, IEEE, Jun Wang, Senior Member, IEEE, and

More information

Independent Component Analysis

Independent Component Analysis 1 Independent Component Analysis Background paper: http://www-stat.stanford.edu/ hastie/papers/ica.pdf 2 ICA Problem X = AS where X is a random p-vector representing multivariate input measurements. S

More information