Fighting the Curse of Dimensionality: Compressive Sensing in Exploration Seismology

Size: px
Start display at page:

Download "Fighting the Curse of Dimensionality: Compressive Sensing in Exploration Seismology"

Transcription

1 Fighting the Curse of Dimensionality: Compressive Sensing in Exploration Seismology Herrmann, F.J.; Friedlander, M.P.; Yilmat, O. Signal Processing Magazine, IEEE, vol.29, no.3, pp Andreas Gaich, Andrea Zabaznoska Advanced Signal Processing 1, Seminar Dec 10, 2012 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 1/68

2 Outline Introduction Mathematical Background of CS The Wavelet-Domain Reflection Seismology Acquisition Schemes Seismic Wavefield Representation Compressive Seismic Computation and Imaging SPGL1 Solver Full Waveform Inversion and PGS Pareto Curves Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 2/68

3 Introduction On Compressive Sensing I Shortcomings of the typical workflows Current seismic techniques rely on massive data volumes as all conducted experiments produce enormous amounts of data. Most of the data consists of reflected energy within a frequency content of roughly [5-100] Hz. Moving into geologically more complex areas of the Earth which makes the correction for wavepaths along which the reflected data traveled impossible (geometry optics approximation of wave propagation). Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 3/68

4 Introduction On Compressive Sensing II Current trends Alternative sampling strategy that leverages recent insights from compressive sensing (CS) towards seismic data acquisition and processing. Introducing CS as a novel nonlinear sampling paradigm, a randomized dimensionality reduction approach effective for acquiring signals that have a sparse representation in some transform domain. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 4/68

5 Introduction On Compressive Sensing III Current trends (cont d) We need a workflow, a framework. We need to come up with sub-nyquist sampling schemes whose sampling is proportional to the sparsity of the problem and not to the dimensionality. Luckily audio, image and seismic signals admit sparse approximations, i.e. they can be well approximated by linear superposition and nonlinear recovery algorithms, such as the l1 norm. l0, l1, l2 norm Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 5/68

6 Introduction On Compressive Sensing VI Compressive sensing (CS) l1 norm Sparsity promotion Wavelets, curvelets, noiselets Exploration seismology Seismic data acquisition Seismic imaging Convex optimization Pareto curves Seismic inversion Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 6/68

7 The Math Behind CS I x R N where x 0 k y = Ψ x... Ψ is an nxn matrix n << N (#msrs.<<ambient dims.) Init. assumptions (1) x... is a high-dimensional signal Ψ... is an full-rank nxn matrix y... non-adaptive linear measurements Goal: obtain x from non-adaptive linear measurements y Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 7/68

8 The Math Behind CS II b = Ψ z where z is sparse minimize z 1 s.t. Ψ z = b Ψ... measurement matrix Sparse recovery problem (2) Goal: find x as the solution of b = Ψ z 1 st Problem: b = Ψ z has infinitely many solutions. 2 nd Problem: Sensitivity to the sparsity assumption. 3 rd Problem: Sensitivity to additive noise, thus not useful in practice. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 8/68

9 The Math Behind CS III Theorem 1 Suppose Ψ is an nxn Gaussian random matrix. If n k log( N n ) then with overwhelming probability we can recover all k-sparse x from y = Ψ x. Problem It is naive to expect signals in practice to be sparse. More realistically, the magnitude of the coefficients decays rapidly and the coefficient vector contains only few significant entries. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 9/68

10 The Math Behind CS IV Theorem 2 b = Ψ z + e, e 2 ɛ minimize z 1 subject to Ψ z b 2 ɛ x... the solution for z (3) Corollary x x 2 C 1 ɛ + C 2 k 1 2 σk (x) n = O(k log N n ) σ k (x)... the best k-term approx. error (4) Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 10/68

11 The Math Behind CS V It has been empirically shown that seismic signals and CS yield small recovery errors, even in scenarios with much degraded sampling ratios. The recovered result is within the noise level and nearly as accurate as the approximation we would obtain by measuring directly the largest k entries of x. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 11/68

12 Solving the Sparse-Optimization Problem The sparsest solution of a severely underdetermined linear system can be recovered exactly by seeking the minimum one-norm (l1 ) solution. The l1 norm finds sparse solutions (whiteboard) Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 12/68

13 Meanwhile in the Wavelet-Domain... I In general seismic signals admit sparse approximations in terms of curvelets: f R N f = S H x and S P XN, P N Init. assumptions b = Ψ f = Ψ S H x (5) f... a compressed data vector S H... the superscript denotes the adjoint Goal: Choose Ψ for a given S s.t. Ψ S H is a good measurement matrix. (whiteboard) Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 13/68

14 Meanwhile in the Wavelet-Domain... II Problem What happens if the sparsifying dictionary is over-complete: the columns of the S H matrix are correlated? there are infinitely many x that explain the same signal f? S P xn... P > N? Solving the sparse optimization problem A universal strategy for choosing Ψ that doesn t require prior knowledge of the sparsity basis S: if we choose Ψ to be appropriate random measurement matrix, then Ψ S H is guaranteed to also be a good measurement matrix independent of S. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 14/68

15 Meanwhile in the Wavelet-Domain... III Bottomline: Many random sensing matrices are universally incoherent with any fixed basis with very high probability. The smaller the coherence between the randomly chosen matrix and the fixed basis, the fewer the samples required. This matches numerical and practical experience. Promote sparsity as a prior via one-norm regularization to overcome the singular nature of S. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 15/68

16 Reflection Seismology - Introduction Proceeding Generatation of Seismic source signal Seismic waves are reflected at the layers of the subsurface Reflected waves are measured by receivers at the ground Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 16/68

17 Reflection Seismology - Introduction - cont d Characteristic Acoustic Impedance Seismic waves travel in Earth at a speed governed by the Acoustic Impedance Z 0 = ρ c Material Property Reflection occurs at the boundary between two materials with different Characteristic Acoustic Impedances R = Z 1 Z 0 Z 1 + Z 0 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 17/68

18 Reflection Seismology - Introduction - cont d Reflection at normal and non-normal incidence Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 18/68

19 Reflection Seismology - Introduction cont d Sources Should ideally only emit P-Waves Impulse f.e. Explotions, Airgun, Earthquakes Source hardly predictible Sweep Vibrator truck Source exactly known Correlation at the receivers necessary Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 19/68

20 Reflection Seismology - Introduction cont d Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 20/68

21 Acquisition Schemes Receiver Land receiver are so called Geophones consider only vertical movement of the earth Common Gather Types Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 21/68

22 Traditional Acquisition Encountered to the traditional Nyquist Theorem to avoid Spatial Aliasing Needs for higher resolution images leads to exponentially increasing costs Acquisition of spatio-temporal wavefield in up to five dimensions f(t, x) L 2 ((0, T ] [ L, L]) T in the order of seconds; L in the order of kilometer Sampling intervalls in the order of milliseconds and meters Dimension reduction needed Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 22/68

23 Compressive Sensing Approach Combines sampling and coding in one single step by a randomized subsampling technique Encoding is linear and does not require access to high-resolution data during encoding Based on: Randomized Sampling Sparsifying Transforms Sparsity-promotion recovery by convex optimization Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 23/68

24 Examples Periodic versus uniformly random subsampling If signals permits a sparse transform-domain representation it suffices to sample at a rate that is lower than Nyquist. Recover signals from far fewer randomly placed samples In Seismology: Use seismic arrays with fewer geophones selected uniformly random from regular sampling grids with spacings defined by Nyquist This turns coherent aliases into Gaussian white noise Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 24/68

25 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 25/68

26 In Geophysical: subsampling-related artifacts are commonly known as spectral leakage Depend on the degree of subsampling Characteristics depend on the irregularity of the sampling Remove noisy artifacts by sparse-recovery procedures Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 26/68

27 Seismic Wavefield Representation Sparse Signal representation concerning CS Leverage structure within signals to reduce sampling Look for transform-domains that concentrate the signals energy in a few number of coefficients Consider transforms that are fast, multiscale and multidirectional Appropriate Transforms Curvelet Transform Wave Atom Transform Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 27/68

28 Wavelet Transformation Generalizes the Fourier transform by using a basis that represents both location and spatial frequency Define orthogonal basis functions as dilations and translations of Mother functions also called analyzing wavelets Φ (sl) (x) = 2 s 2 Φ(2 s x l) Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 28/68

29 Wavelet Transformation - cont d Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 29/68

30 Wavelet Transformation - cont d Wavelets do not utilize geometric properties of wavefields Curved structures as superposition of Multiscale Fat Dots Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 30/68

31 Curvelet Transformation Extension of wavelets by additional Orientation in Localization ( ) ϕ j,l,k (x) = ϕ j R Θl (x x (j,l) k ) [ cos Θl sin Θ R Θl = l sin Θ l cos Θ l ], x (j,l) k = R 1 Θ l (k 1 2 j, k 2 2 j/2 ) Θ l = 2π 2 j/2 l, l = 0, 1,..., 0 Θ l < 2π Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 31/68

32 Curvelet Transformation - cont d Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 32/68

33 Curvelets Waveatoms Both obey a so called parabolic scaling Curvelets: needle-like shapes Wave Atoms: Oscillatory patterns Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 33/68

34 Performance measure for transforms Approximation Error real signals are not strictly sparse but transfer-domain coefficients often decay rapidly Orthonormal basis: Decay rate directly linked to the decay of the nonlinear approximation error σ(k) = f f k 2 This does not hold for redundant transforms Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 34/68

35 Alternative Approximation Error Based on solution of a sparsity promoting program min x 1 subject to S H x = f To account for different redundancies plot SNR as a function of the sparsity ratio ρ = k/p SNR(ρ) = 20 log f f ρ f Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 35/68

36 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 36/68

37 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 37/68

38 Subsampling of shots Aim Breaking the periodicity of coherent sampling Oppurtunities Selections of subsets of sources Design of incoherent simultaneous-source experiments Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 38/68

39 Subsampling of shots - cont d Measurement basises Sequential-source acquisition I = I Ns I Nt Simultaneous-source acquisition Incomplete data M = G Ns I Nt R = R ns I Nt n s N s Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 39/68

40 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 40/68

41 Subsampling of shots cont d Coherence depends on the interplay between restriction, measurement and synthesis matrices Another performance measure where δ = n/n, SNR(δ) = 20 log f f δ f f δ = S H x δ, x δ = argmin x 1 subject to A δ x = b A δ = R δ M δ S H Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 41/68

42 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 42/68

43 Compressive Seismic Computation Overview Compressive Simulation Compressive Imaging Compressive Inversion SPGL1 - Spectral Projected-Gradient for L1 norm Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 43/68

44 Compressive Simulation Aim: Simulation of P-Waves Traditionally solved with the time-harmonic Helmholtz equation large linear system of PDEs that discretizes the underlying wave equation Use linearity in the sources to reduce the number of sequential shots into a small number of supershots complexity reduction from O(n 4 ) to O(n 3 log n) Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 44/68

45 Compressive Imaging Motivation Locate mineral and oil sources Building Industry Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 45/68

46 The Seismic Image Problem Requires inversion of the linearized time-harmonic Born-scattering matrix Seismic data is decomposed through Fourier transform into monochromatic wavefields minimize b Ax 2 2 = K b i A i x 2 2 i=1 b C N f N rn s, x R M, K = N f N s Each iteration needs solutions of a big set of PDEs Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 46/68

47 Solution by batching Take Mini-Batches with K K monochromatic supershots Solve reduced system minimize b Ax 2 2 = b j A j x 2 2 K i=1 b j = K w ij b i, A j = i=1 K w ij A i i=1 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 47/68

48 Solution by batching - cont d Sub-selection and mixing in the source-frequency space: RM(b Ax) 2 2 R = R Σ I R Ω, M = M Σ I I R Σ R n s n s, R Ω R n f n f, M R (KNr( (KNr) Solution reduced by a factor of K /K Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 48/68

49 Solution by sparsity promotion Averaging is not able to remove the source crosstalk efficiently use transform-domain sparsity promotion Additionaly: Redraw supershots at each iteration Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 49/68

50 Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 50/68

51 SPGL1 Solver large-scale sparse reconstruction solver Basis pursuit minimize x 1 subject to Ax = b Basis pursuit denoise minimize x 1 subject to Ax b 2 σ Lasso minimize Ax b 2 subject to x 1 τ Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 51/68

52 SPGL Solver - cont d Basis pursuit denoise Solve BP σ by a sequence of subproblems LS τ At each iteration k refine τ k such that τ k τ σ Derive sequence of estimates τ k by applying Newton s method to find the root of value function φ(τ) = σ Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 52/68

53 SPGL Solver - cont d LS τ solved by Spectral projected gradient P τ [b] := {argmin b x 2 subject to x 1 τ} Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 53/68

54 SPGL Solver - cont d Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 54/68

55 Full Waveform Inversion FWI I Most industry practice is based on a geometric optics approximation of the wave propagation. A smooth velocity model that describes the propagation speed of the waves in the subsurface (lack of knowledge). FWI relies on modeling the data by solving the wave equation (PDE), and adapting the model parameters in order to minimize the data misfit based on the LS-optimization. Minimization of the data misfit requires low wavenumbers. And yes, this is now becoming possible! Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 55/68

56 Full Waveform Inversion FWI II Ghost-free seismic data Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 56/68

57 Full Waveform Inversion FWI III More Wavenumbers Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 57/68

58 Full Waveform Inversion FWI IV More Wavenumbers (cont d) Broadband frequency signals propagating through the earth and the ocean. The wavenumber of any seismic signal is related in a rather complex manner to its seismic frequency, the seismic velocity and the offset between source and receiver locations, recorded along a towed streamer. Removing the receiver ghost and deep streamer towing allows recording data in a low noise environment. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 58/68

59 Full Waveform Inversion FWI V More Wavenumbers (cont d) Wavefield separation during processing allows building high resolution velocity models and high resolution imaging. LOW WAVENUMBERS: data rich in low frequencies. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 59/68

60 Full Waveform Inversion FWI VI Wavefield Separation Two interfering wavefields: up-going wavefield scattered upwards from the earth and the time-delayed version that is reflected downwards from the sea surface. Key differentiation technique of GeoStreamer. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 60/68

61 Full Waveform Inversion FWI VII Underdetermined system of linear equations: f = S H x A = RMS H Ax = b minimize x 0 s.t. Ax = b l 0... #of nonzero elements Basic problem (6) This cannot be approximated by a convex approximation problem. One of the major findings of CS is that under some conditions on A and x, the solution can be recovered by solving the convex optimization problem. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 61/68

62 Full Waveform Inversion FWI VIII Basis Pursuit (BP): minimize x 1 s.t. Ax = b (7) Recovers the correct sparse signal depending on the sparsity level of x, the #of measurements and the restricted isometry property (RIP) of A. The RIP constant measures how far the matrix A is from a unitary matrix when acting on sparse vectors (A A = I). The columns of an unitary matrix form an orthonormal basis, as well as the rows. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 62/68

63 Full Waveform Inversion FWI IX Quadratic Programming (QP): minimize 1 2 Ax b λ x 1 (8) Basis Pursuit Denoise Problem (BPDP): minimize x 1 s.t. b Ax 2 σ (9) LASSO (LS): min 1 2 b Ax 2 s.t. x 1 τ (10) For each σ, there are unique values of λ and τ so that the solutions for QP and LASSO coincide. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 63/68

64 Full Waveform Inversion FWI X LASSO (cont d) The estimate of the one norm of the solution (τ) is seldom available for geophysical problems, but is a key internal problem used by Spectral Projected Gradient Method (SPGL). The SPGL can be used to quickly solve the LS equation for very large linear systems and it has already been proven to be very successful in solving large-scale CS problems in seismic exploration. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 64/68

65 Pareto Curves I Pareto Curves: Trace the optimal trade-off between the data misfit and some prior model. Are commonly used in problems with two-norm priors and recently explored in terms of one-norm regularization. Are convex and decreasing, i.e. regular. This regularity means it is possible to obtain a good approximation to the curve with very few interpolation points. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 65/68

66 Pareto Curves II Pareto Curves (cont d) Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 66/68

67 References I F.J. Herrmann, M.P. Friedlander, O. Yilmaz, Fighting the curse of dimensionality: compressive sensing in exploration seismology, IEEE Signal Processing Magazine, vol. 29, no. 3, pp , E.J. Candes, M.B. Wakin, An Introduction to Compressive Sampling, IEEE Signal Processing Magazine, vol. 25, no. 2, pp , R. Baraniuk, Compressive Sensing, Lecture Notes in IEEE Signal Processing Magazine, vol. 24, pp A. Aravkin, X. Li, F.J. Hermann, Fast Seismic Imaging For Marine Data, The University of British Columbia Technical Report., October J. Ma, G. Plonka, The Curvelet Transform, IEEE Signal Processing Magazine, vol. 27, no. 2, pp , March J. Ma, Characterization of textural surfaces using wave atoms, Applied Physics Letters, vol. 90, no. 26, June Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 67/68

68 References II E.v.d. Berg, M.P. Friedlander, Probing the Pareto Frontier for Basis Pursuit Solutions, The University of British Columbia Technical Report., May E. Hager, Full Azimuth Seismic Acquisition with Coil Shooting, 8. Biennial International Conference and Exposition on Petroleum Geophysics Hyderabad, F.J. Herrmann, X. Li, A.Y. Aravkin, T. Leeuwen, A modified, sparsity promoting, Gauss-Newton algorithm for seismic waveform inversion, Dept. of Earth and Ocean Sciences, University of British Columbia. F.J. Herrmann, Randomized sampling and sparsity: getting more information from fewer samples, The University of British Columbia Technical Report, TR , G. Hennenfent, E.v.d. Berg, M.P. Friedlander, F.J. Herrmann, New Insights into one-norm solvers from the Pareto curve, Geophysics, vol. 73, no. 4, pp. A23 A26. Andreas Gaich, Andrea Zabaznoska Dec 10, 2012 page 68/68

FWI with Compressive Updates Aleksandr Aravkin, Felix Herrmann, Tristan van Leeuwen, Xiang Li, James Burke

FWI with Compressive Updates Aleksandr Aravkin, Felix Herrmann, Tristan van Leeuwen, Xiang Li, James Burke Consortium 2010 FWI with Compressive Updates Aleksandr Aravkin, Felix Herrmann, Tristan van Leeuwen, Xiang Li, James Burke SLIM University of British Columbia Full Waveform Inversion The Full Waveform

More information

Seismic data interpolation and denoising using SVD-free low-rank matrix factorization

Seismic data interpolation and denoising using SVD-free low-rank matrix factorization Seismic data interpolation and denoising using SVD-free low-rank matrix factorization R. Kumar, A.Y. Aravkin,, H. Mansour,, B. Recht and F.J. Herrmann Dept. of Earth and Ocean sciences, University of British

More information

Compressive Sensing Applied to Full-wave Form Inversion

Compressive Sensing Applied to Full-wave Form Inversion Compressive Sensing Applied to Full-wave Form Inversion Felix J. Herrmann* fherrmann@eos.ubc.ca Joint work with Yogi Erlangga, and Tim Lin *Seismic Laboratory for Imaging & Modeling Department of Earth

More information

Fighting the curse of dimensionality: compressive sensing in exploration seismology

Fighting the curse of dimensionality: compressive sensing in exploration seismology Fighting the curse of dimensionality: compressive sensing in exploration seismology Felix J. Herrmann, Michael P. Friedlander, Özgür Yılmaz January 2, 202 Abstract Many seismic exploration techniques rely

More information

Accelerated large-scale inversion with message passing Felix J. Herrmann, the University of British Columbia, Canada

Accelerated large-scale inversion with message passing Felix J. Herrmann, the University of British Columbia, Canada Accelerated large-scale inversion with message passing Felix J. Herrmann, the University of British Columbia, Canada SUMMARY To meet current-day challenges, exploration seismology increasingly relies on

More information

Compressive sampling meets seismic imaging

Compressive sampling meets seismic imaging Compressive sampling meets seismic imaging Felix J. Herrmann fherrmann@eos.ubc.ca http://slim.eos.ubc.ca joint work with Tim Lin and Yogi Erlangga Seismic Laboratory for Imaging & Modeling Department of

More information

Seismic wavefield inversion with curvelet-domain sparsity promotion

Seismic wavefield inversion with curvelet-domain sparsity promotion Seismic wavefield inversion with curvelet-domain sparsity promotion Felix J. Herrmann* fherrmann@eos.ubc.ca Deli Wang** wangdeli@email.jlu.edu.cn *Seismic Laboratory for Imaging & Modeling Department of

More information

Introduction How it works Theory behind Compressed Sensing. Compressed Sensing. Huichao Xue. CS3750 Fall 2011

Introduction How it works Theory behind Compressed Sensing. Compressed Sensing. Huichao Xue. CS3750 Fall 2011 Compressed Sensing Huichao Xue CS3750 Fall 2011 Table of Contents Introduction From News Reports Abstract Definition How it works A review of L 1 norm The Algorithm Backgrounds for underdetermined linear

More information

Low-rank Promoting Transformations and Tensor Interpolation - Applications to Seismic Data Denoising

Low-rank Promoting Transformations and Tensor Interpolation - Applications to Seismic Data Denoising Low-rank Promoting Transformations and Tensor Interpolation - Applications to Seismic Data Denoising Curt Da Silva and Felix J. Herrmann 2 Dept. of Mathematics 2 Dept. of Earth and Ocean Sciences, University

More information

Source estimation for frequency-domain FWI with robust penalties

Source estimation for frequency-domain FWI with robust penalties Source estimation for frequency-domain FWI with robust penalties Aleksandr Y. Aravkin, Tristan van Leeuwen, Henri Calandra, and Felix J. Herrmann Dept. of Earth and Ocean sciences University of British

More information

Generalized Orthogonal Matching Pursuit- A Review and Some

Generalized Orthogonal Matching Pursuit- A Review and Some Generalized Orthogonal Matching Pursuit- A Review and Some New Results Department of Electronics and Electrical Communication Engineering Indian Institute of Technology, Kharagpur, INDIA Table of Contents

More information

Introduction to Compressed Sensing

Introduction to Compressed Sensing Introduction to Compressed Sensing Alejandro Parada, Gonzalo Arce University of Delaware August 25, 2016 Motivation: Classical Sampling 1 Motivation: Classical Sampling Issues Some applications Radar Spectral

More information

Compressed sensing. Or: the equation Ax = b, revisited. Terence Tao. Mahler Lecture Series. University of California, Los Angeles

Compressed sensing. Or: the equation Ax = b, revisited. Terence Tao. Mahler Lecture Series. University of California, Los Angeles Or: the equation Ax = b, revisited University of California, Los Angeles Mahler Lecture Series Acquiring signals Many types of real-world signals (e.g. sound, images, video) can be viewed as an n-dimensional

More information

Lecture 22: More On Compressed Sensing

Lecture 22: More On Compressed Sensing Lecture 22: More On Compressed Sensing Scribed by Eric Lee, Chengrun Yang, and Sebastian Ament Nov. 2, 207 Recap and Introduction Basis pursuit was the method of recovering the sparsest solution to an

More information

Matrix Probing and Simultaneous Sources: A New Approach for Preconditioning the Hessian

Matrix Probing and Simultaneous Sources: A New Approach for Preconditioning the Hessian Matrix Probing and Simultaneous Sources: A New Approach for Preconditioning the Hessian Curt Da Silva 1 and Felix J. Herrmann 2 1 Dept. of Mathematics 2 Dept. of Earth and Ocean SciencesUniversity of British

More information

AN INTRODUCTION TO COMPRESSIVE SENSING

AN INTRODUCTION TO COMPRESSIVE SENSING AN INTRODUCTION TO COMPRESSIVE SENSING Rodrigo B. Platte School of Mathematical and Statistical Sciences APM/EEE598 Reverse Engineering of Complex Dynamical Networks OUTLINE 1 INTRODUCTION 2 INCOHERENCE

More information

3D INTERPOLATION USING HANKEL TENSOR COMPLETION BY ORTHOGONAL MATCHING PURSUIT A. Adamo, P. Mazzucchelli Aresys, Milano, Italy

3D INTERPOLATION USING HANKEL TENSOR COMPLETION BY ORTHOGONAL MATCHING PURSUIT A. Adamo, P. Mazzucchelli Aresys, Milano, Italy 3D INTERPOLATION USING HANKEL TENSOR COMPLETION BY ORTHOGONAL MATCHING PURSUIT A. Adamo, P. Mazzucchelli Aresys, Milano, Italy Introduction. Seismic data are often sparsely or irregularly sampled along

More information

COMPRESSED SENSING IN PYTHON

COMPRESSED SENSING IN PYTHON COMPRESSED SENSING IN PYTHON Sercan Yıldız syildiz@samsi.info February 27, 2017 OUTLINE A BRIEF INTRODUCTION TO COMPRESSED SENSING A BRIEF INTRODUCTION TO CVXOPT EXAMPLES A Brief Introduction to Compressed

More information

Structured tensor missing-trace interpolation in the Hierarchical Tucker format Curt Da Silva and Felix J. Herrmann Sept. 26, 2013

Structured tensor missing-trace interpolation in the Hierarchical Tucker format Curt Da Silva and Felix J. Herrmann Sept. 26, 2013 Structured tensor missing-trace interpolation in the Hierarchical Tucker format Curt Da Silva and Felix J. Herrmann Sept. 6, 13 SLIM University of British Columbia Motivation 3D seismic experiments - 5D

More information

Time domain sparsity promoting LSRTM with source estimation

Time domain sparsity promoting LSRTM with source estimation Time domain sparsity promoting LSRTM with source estimation Mengmeng Yang, Philipp Witte, Zhilong Fang & Felix J. Herrmann SLIM University of British Columbia Motivation Features of RTM: pros - no dip

More information

An Introduction to Sparse Approximation

An Introduction to Sparse Approximation An Introduction to Sparse Approximation Anna C. Gilbert Department of Mathematics University of Michigan Basic image/signal/data compression: transform coding Approximate signals sparsely Compress images,

More information

SPARSE signal representations have gained popularity in recent

SPARSE signal representations have gained popularity in recent 6958 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 57, NO. 10, OCTOBER 2011 Blind Compressed Sensing Sivan Gleichman and Yonina C. Eldar, Senior Member, IEEE Abstract The fundamental principle underlying

More information

Sparsity-promoting migration with multiples

Sparsity-promoting migration with multiples Sparsity-promoting migration with multiples Tim Lin, Ning Tu and Felix Herrmann SLIM Seismic Laboratory for Imaging and Modeling the University of British Columbia Courtesy of Verschuur, 29 SLIM Motivation..

More information

Near Ideal Behavior of a Modified Elastic Net Algorithm in Compressed Sensing

Near Ideal Behavior of a Modified Elastic Net Algorithm in Compressed Sensing Near Ideal Behavior of a Modified Elastic Net Algorithm in Compressed Sensing M. Vidyasagar Cecil & Ida Green Chair The University of Texas at Dallas M.Vidyasagar@utdallas.edu www.utdallas.edu/ m.vidyasagar

More information

Lecture Notes 9: Constrained Optimization

Lecture Notes 9: Constrained Optimization Optimization-based data analysis Fall 017 Lecture Notes 9: Constrained Optimization 1 Compressed sensing 1.1 Underdetermined linear inverse problems Linear inverse problems model measurements of the form

More information

Compressive simultaneous full-waveform simulation

Compressive simultaneous full-waveform simulation Compressive simultaneous full-waveform simulation Felix J. Herrmann 1, Yogi Erlangga 1 and Tim T. Y. Lin 1 (August 14, 2008) Running head: compressive simulations ABSTRACT The fact that the numerical complexity

More information

Sparse linear models

Sparse linear models Sparse linear models Optimization-Based Data Analysis http://www.cims.nyu.edu/~cfgranda/pages/obda_spring16 Carlos Fernandez-Granda 2/22/2016 Introduction Linear transforms Frequency representation Short-time

More information

Compressed Sensing Using Reed- Solomon and Q-Ary LDPC Codes

Compressed Sensing Using Reed- Solomon and Q-Ary LDPC Codes Compressed Sensing Using Reed- Solomon and Q-Ary LDPC Codes Item Type text; Proceedings Authors Jagiello, Kristin M. Publisher International Foundation for Telemetering Journal International Telemetering

More information

Sparse Solutions of an Undetermined Linear System

Sparse Solutions of an Undetermined Linear System 1 Sparse Solutions of an Undetermined Linear System Maddullah Almerdasy New York University Tandon School of Engineering arxiv:1702.07096v1 [math.oc] 23 Feb 2017 Abstract This work proposes a research

More information

SUMMARY. H (ω)u(ω,x s ;x) :=

SUMMARY. H (ω)u(ω,x s ;x) := Interpolating solutions of the Helmholtz equation with compressed sensing Tim TY Lin*, Evgeniy Lebed, Yogi A Erlangga, and Felix J Herrmann, University of British Columbia, EOS SUMMARY We present an algorithm

More information

Optimization for Compressed Sensing

Optimization for Compressed Sensing Optimization for Compressed Sensing Robert J. Vanderbei 2014 March 21 Dept. of Industrial & Systems Engineering University of Florida http://www.princeton.edu/ rvdb Lasso Regression The problem is to solve

More information

Making Flippy Floppy

Making Flippy Floppy Making Flippy Floppy James V. Burke UW Mathematics jvburke@uw.edu Aleksandr Y. Aravkin IBM, T.J.Watson Research sasha.aravkin@gmail.com Michael P. Friedlander UBC Computer Science mpf@cs.ubc.ca Current

More information

Recovering overcomplete sparse representations from structured sensing

Recovering overcomplete sparse representations from structured sensing Recovering overcomplete sparse representations from structured sensing Deanna Needell Claremont McKenna College Feb. 2015 Support: Alfred P. Sloan Foundation and NSF CAREER #1348721. Joint work with Felix

More information

Compressive Sensing and Beyond

Compressive Sensing and Beyond Compressive Sensing and Beyond Sohail Bahmani Gerorgia Tech. Signal Processing Compressed Sensing Signal Models Classics: bandlimited The Sampling Theorem Any signal with bandwidth B can be recovered

More information

PHASE RETRIEVAL OF SPARSE SIGNALS FROM MAGNITUDE INFORMATION. A Thesis MELTEM APAYDIN

PHASE RETRIEVAL OF SPARSE SIGNALS FROM MAGNITUDE INFORMATION. A Thesis MELTEM APAYDIN PHASE RETRIEVAL OF SPARSE SIGNALS FROM MAGNITUDE INFORMATION A Thesis by MELTEM APAYDIN Submitted to the Office of Graduate and Professional Studies of Texas A&M University in partial fulfillment of the

More information

Lecture: Introduction to Compressed Sensing Sparse Recovery Guarantees

Lecture: Introduction to Compressed Sensing Sparse Recovery Guarantees Lecture: Introduction to Compressed Sensing Sparse Recovery Guarantees http://bicmr.pku.edu.cn/~wenzw/bigdata2018.html Acknowledgement: this slides is based on Prof. Emmanuel Candes and Prof. Wotao Yin

More information

Sparsity Models. Tong Zhang. Rutgers University. T. Zhang (Rutgers) Sparsity Models 1 / 28

Sparsity Models. Tong Zhang. Rutgers University. T. Zhang (Rutgers) Sparsity Models 1 / 28 Sparsity Models Tong Zhang Rutgers University T. Zhang (Rutgers) Sparsity Models 1 / 28 Topics Standard sparse regression model algorithms: convex relaxation and greedy algorithm sparse recovery analysis:

More information

Compressive-wavefield simulations

Compressive-wavefield simulations Compressive-wavefield simulations Felix Herrmann Yogi Erlangga, Tim. Lin To cite this version: Felix Herrmann Yogi Erlangga, Tim. Lin. Compressive-wavefield simulations. Laurent Fesquet and Bruno Torrésani.

More information

ECE 8201: Low-dimensional Signal Models for High-dimensional Data Analysis

ECE 8201: Low-dimensional Signal Models for High-dimensional Data Analysis ECE 8201: Low-dimensional Signal Models for High-dimensional Data Analysis Lecture 7: Matrix completion Yuejie Chi The Ohio State University Page 1 Reference Guaranteed Minimum-Rank Solutions of Linear

More information

ECE 8201: Low-dimensional Signal Models for High-dimensional Data Analysis

ECE 8201: Low-dimensional Signal Models for High-dimensional Data Analysis ECE 8201: Low-dimensional Signal Models for High-dimensional Data Analysis Lecture 3: Sparse signal recovery: A RIPless analysis of l 1 minimization Yuejie Chi The Ohio State University Page 1 Outline

More information

Sparsity Regularization

Sparsity Regularization Sparsity Regularization Bangti Jin Course Inverse Problems & Imaging 1 / 41 Outline 1 Motivation: sparsity? 2 Mathematical preliminaries 3 l 1 solvers 2 / 41 problem setup finite-dimensional formulation

More information

Sparse Legendre expansions via l 1 minimization

Sparse Legendre expansions via l 1 minimization Sparse Legendre expansions via l 1 minimization Rachel Ward, Courant Institute, NYU Joint work with Holger Rauhut, Hausdorff Center for Mathematics, Bonn, Germany. June 8, 2010 Outline Sparse recovery

More information

Optimal Value Function Methods in Numerical Optimization Level Set Methods

Optimal Value Function Methods in Numerical Optimization Level Set Methods Optimal Value Function Methods in Numerical Optimization Level Set Methods James V Burke Mathematics, University of Washington, (jvburke@uw.edu) Joint work with Aravkin (UW), Drusvyatskiy (UW), Friedlander

More information

Compressive Sensing (CS)

Compressive Sensing (CS) Compressive Sensing (CS) Luminita Vese & Ming Yan lvese@math.ucla.edu yanm@math.ucla.edu Department of Mathematics University of California, Los Angeles The UCLA Advanced Neuroimaging Summer Program (2014)

More information

Uncertainty quantification for Wavefield Reconstruction Inversion

Uncertainty quantification for Wavefield Reconstruction Inversion Uncertainty quantification for Wavefield Reconstruction Inversion Zhilong Fang *, Chia Ying Lee, Curt Da Silva *, Felix J. Herrmann *, and Rachel Kuske * Seismic Laboratory for Imaging and Modeling (SLIM),

More information

Conditions for Robust Principal Component Analysis

Conditions for Robust Principal Component Analysis Rose-Hulman Undergraduate Mathematics Journal Volume 12 Issue 2 Article 9 Conditions for Robust Principal Component Analysis Michael Hornstein Stanford University, mdhornstein@gmail.com Follow this and

More information

Sensing systems limited by constraints: physical size, time, cost, energy

Sensing systems limited by constraints: physical size, time, cost, energy Rebecca Willett Sensing systems limited by constraints: physical size, time, cost, energy Reduce the number of measurements needed for reconstruction Higher accuracy data subject to constraints Original

More information

Compressed sensing techniques for hyperspectral image recovery

Compressed sensing techniques for hyperspectral image recovery Compressed sensing techniques for hyperspectral image recovery A. Abrardo, M. Barni, C. M. Carretti, E. Magli, S. Kuiteing Kamdem, R. Vitulli ABSTRACT Compressed Sensing (CS) theory is progressively gaining

More information

Compressed Sensing: Extending CLEAN and NNLS

Compressed Sensing: Extending CLEAN and NNLS Compressed Sensing: Extending CLEAN and NNLS Ludwig Schwardt SKA South Africa (KAT Project) Calibration & Imaging Workshop Socorro, NM, USA 31 March 2009 Outline 1 Compressed Sensing (CS) Introduction

More information

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation Instructor: Moritz Hardt Email: hardt+ee227c@berkeley.edu Graduate Instructor: Max Simchowitz Email: msimchow+ee227c@berkeley.edu

More information

Multiscale Geometric Analysis: Thoughts and Applications (a summary)

Multiscale Geometric Analysis: Thoughts and Applications (a summary) Multiscale Geometric Analysis: Thoughts and Applications (a summary) Anestis Antoniadis, University Joseph Fourier Assimage 2005,Chamrousse, February 2005 Classical Multiscale Analysis Wavelets: Enormous

More information

Full-Waveform Inversion with Gauss- Newton-Krylov Method

Full-Waveform Inversion with Gauss- Newton-Krylov Method Full-Waveform Inversion with Gauss- Newton-Krylov Method Yogi A. Erlangga and Felix J. Herrmann {yerlangga,fherrmann}@eos.ubc.ca Seismic Laboratory for Imaging and Modeling The University of British Columbia

More information

Compressibility of Infinite Sequences and its Interplay with Compressed Sensing Recovery

Compressibility of Infinite Sequences and its Interplay with Compressed Sensing Recovery Compressibility of Infinite Sequences and its Interplay with Compressed Sensing Recovery Jorge F. Silva and Eduardo Pavez Department of Electrical Engineering Information and Decision Systems Group Universidad

More information

Vollständige Inversion seismischer Wellenfelder - Erderkundung im oberflächennahen Bereich

Vollständige Inversion seismischer Wellenfelder - Erderkundung im oberflächennahen Bereich Seminar des FA Ultraschallprüfung Vortrag 1 More info about this article: http://www.ndt.net/?id=20944 Vollständige Inversion seismischer Wellenfelder - Erderkundung im oberflächennahen Bereich Thomas

More information

Random projections. 1 Introduction. 2 Dimensionality reduction. Lecture notes 5 February 29, 2016

Random projections. 1 Introduction. 2 Dimensionality reduction. Lecture notes 5 February 29, 2016 Lecture notes 5 February 9, 016 1 Introduction Random projections Random projections are a useful tool in the analysis and processing of high-dimensional data. We will analyze two applications that use

More information

New Coherence and RIP Analysis for Weak. Orthogonal Matching Pursuit

New Coherence and RIP Analysis for Weak. Orthogonal Matching Pursuit New Coherence and RIP Analysis for Wea 1 Orthogonal Matching Pursuit Mingrui Yang, Member, IEEE, and Fran de Hoog arxiv:1405.3354v1 [cs.it] 14 May 2014 Abstract In this paper we define a new coherence

More information

Wavefield Reconstruction Inversion (WRI) a new take on wave-equation based inversion Felix J. Herrmann

Wavefield Reconstruction Inversion (WRI) a new take on wave-equation based inversion Felix J. Herrmann Wavefield Reconstruction Inversion (WRI) a new take on wave-equation based inversion Felix J. Herrmann SLIM University of British Columbia van Leeuwen, T and Herrmann, F J (2013). Mitigating local minima

More information

Sparse Signal Recovery

Sparse Signal Recovery Sparse Signal Recovery Analysis and Synthesis Formulations with Prior Support Information by Brock Edward Hargreaves B.Sc. (Hons.), The University of Calgary, 2006 A THESIS SUBMITTED IN PARTIAL FULFILLMENT

More information

Phase Transition Phenomenon in Sparse Approximation

Phase Transition Phenomenon in Sparse Approximation Phase Transition Phenomenon in Sparse Approximation University of Utah/Edinburgh L1 Approximation: May 17 st 2008 Convex polytopes Counting faces Sparse Representations via l 1 Regularization Underdetermined

More information

The Pros and Cons of Compressive Sensing

The Pros and Cons of Compressive Sensing The Pros and Cons of Compressive Sensing Mark A. Davenport Stanford University Department of Statistics Compressive Sensing Replace samples with general linear measurements measurements sampled signal

More information

Lecture Notes 5: Multiresolution Analysis

Lecture Notes 5: Multiresolution Analysis Optimization-based data analysis Fall 2017 Lecture Notes 5: Multiresolution Analysis 1 Frames A frame is a generalization of an orthonormal basis. The inner products between the vectors in a frame and

More information

Signal Recovery, Uncertainty Relations, and Minkowski Dimension

Signal Recovery, Uncertainty Relations, and Minkowski Dimension Signal Recovery, Uncertainty Relations, and Minkowski Dimension Helmut Bőlcskei ETH Zurich December 2013 Joint work with C. Aubel, P. Kuppinger, G. Pope, E. Riegler, D. Stotz, and C. Studer Aim of this

More information

Compressed Sensing: Lecture I. Ronald DeVore

Compressed Sensing: Lecture I. Ronald DeVore Compressed Sensing: Lecture I Ronald DeVore Motivation Compressed Sensing is a new paradigm for signal/image/function acquisition Motivation Compressed Sensing is a new paradigm for signal/image/function

More information

Lecture 5 : Sparse Models

Lecture 5 : Sparse Models Lecture 5 : Sparse Models Homework 3 discussion (Nima) Sparse Models Lecture - Reading : Murphy, Chapter 13.1, 13.3, 13.6.1 - Reading : Peter Knee, Chapter 2 Paolo Gabriel (TA) : Neural Brain Control After

More information

Seismic Imaging. William W. Symes. C. I. M. E. Martina Franca September

Seismic Imaging. William W. Symes. C. I. M. E. Martina Franca September Seismic Imaging William W. Symes C. I. M. E. Martina Franca September 2002 www.trip.caam.rice.edu 1 0 offset (km) -4-3 -2-1 1 2 time (s) 3 4 5 How do you turn lots of this... (field seismogram from the

More information

Large-Scale L1-Related Minimization in Compressive Sensing and Beyond

Large-Scale L1-Related Minimization in Compressive Sensing and Beyond Large-Scale L1-Related Minimization in Compressive Sensing and Beyond Yin Zhang Department of Computational and Applied Mathematics Rice University, Houston, Texas, U.S.A. Arizona State University March

More information

2D X-Ray Tomographic Reconstruction From Few Projections

2D X-Ray Tomographic Reconstruction From Few Projections 2D X-Ray Tomographic Reconstruction From Few Projections Application of Compressed Sensing Theory CEA-LID, Thalès, UJF 6 octobre 2009 Introduction Plan 1 Introduction 2 Overview of Compressed Sensing Theory

More information

Thresholds for the Recovery of Sparse Solutions via L1 Minimization

Thresholds for the Recovery of Sparse Solutions via L1 Minimization Thresholds for the Recovery of Sparse Solutions via L Minimization David L. Donoho Department of Statistics Stanford University 39 Serra Mall, Sequoia Hall Stanford, CA 9435-465 Email: donoho@stanford.edu

More information

Exponential decay of reconstruction error from binary measurements of sparse signals

Exponential decay of reconstruction error from binary measurements of sparse signals Exponential decay of reconstruction error from binary measurements of sparse signals Deanna Needell Joint work with R. Baraniuk, S. Foucart, Y. Plan, and M. Wootters Outline Introduction Mathematical Formulation

More information

Beyond incoherence and beyond sparsity: compressed sensing in the real world

Beyond incoherence and beyond sparsity: compressed sensing in the real world Beyond incoherence and beyond sparsity: compressed sensing in the real world Clarice Poon 1st November 2013 University of Cambridge, UK Applied Functional and Harmonic Analysis Group Head of Group Anders

More information

Sparse Optimization Lecture: Sparse Recovery Guarantees

Sparse Optimization Lecture: Sparse Recovery Guarantees Those who complete this lecture will know Sparse Optimization Lecture: Sparse Recovery Guarantees Sparse Optimization Lecture: Sparse Recovery Guarantees Instructor: Wotao Yin Department of Mathematics,

More information

Compressed sensing for radio interferometry: spread spectrum imaging techniques

Compressed sensing for radio interferometry: spread spectrum imaging techniques Compressed sensing for radio interferometry: spread spectrum imaging techniques Y. Wiaux a,b, G. Puy a, Y. Boursier a and P. Vandergheynst a a Institute of Electrical Engineering, Ecole Polytechnique Fédérale

More information

Message Passing Algorithms for Compressed Sensing: II. Analysis and Validation

Message Passing Algorithms for Compressed Sensing: II. Analysis and Validation Message Passing Algorithms for Compressed Sensing: II. Analysis and Validation David L. Donoho Department of Statistics Arian Maleki Department of Electrical Engineering Andrea Montanari Department of

More information

Recovery of Sparse Signals from Noisy Measurements Using an l p -Regularized Least-Squares Algorithm

Recovery of Sparse Signals from Noisy Measurements Using an l p -Regularized Least-Squares Algorithm Recovery of Sparse Signals from Noisy Measurements Using an l p -Regularized Least-Squares Algorithm J. K. Pant, W.-S. Lu, and A. Antoniou University of Victoria August 25, 2011 Compressive Sensing 1 University

More information

Compressed Sensing and Sparse Recovery

Compressed Sensing and Sparse Recovery ELE 538B: Sparsity, Structure and Inference Compressed Sensing and Sparse Recovery Yuxin Chen Princeton University, Spring 217 Outline Restricted isometry property (RIP) A RIPless theory Compressed sensing

More information

Extended Reconstruction Approaches for Saturation Measurements Using Reserved Quantization Indices Li, Peng; Arildsen, Thomas; Larsen, Torben

Extended Reconstruction Approaches for Saturation Measurements Using Reserved Quantization Indices Li, Peng; Arildsen, Thomas; Larsen, Torben Aalborg Universitet Extended Reconstruction Approaches for Saturation Measurements Using Reserved Quantization Indices Li, Peng; Arildsen, Thomas; Larsen, Torben Published in: Proceedings of the 12th IEEE

More information

IEEE SIGNAL PROCESSING LETTERS, VOL. 22, NO. 9, SEPTEMBER

IEEE SIGNAL PROCESSING LETTERS, VOL. 22, NO. 9, SEPTEMBER IEEE SIGNAL PROCESSING LETTERS, VOL. 22, NO. 9, SEPTEMBER 2015 1239 Preconditioning for Underdetermined Linear Systems with Sparse Solutions Evaggelia Tsiligianni, StudentMember,IEEE, Lisimachos P. Kondi,

More information

Making Flippy Floppy

Making Flippy Floppy Making Flippy Floppy James V. Burke UW Mathematics jvburke@uw.edu Aleksandr Y. Aravkin IBM, T.J.Watson Research sasha.aravkin@gmail.com Michael P. Friedlander UBC Computer Science mpf@cs.ubc.ca Vietnam

More information

Sparse & Redundant Signal Representation, and its Role in Image Processing

Sparse & Redundant Signal Representation, and its Role in Image Processing Sparse & Redundant Signal Representation, and its Role in Michael Elad The CS Department The Technion Israel Institute of technology Haifa 3000, Israel Wave 006 Wavelet and Applications Ecole Polytechnique

More information

Low-Dimensional Signal Models in Compressive Sensing

Low-Dimensional Signal Models in Compressive Sensing University of Colorado, Boulder CU Scholar Electrical, Computer & Energy Engineering Graduate Theses & Dissertations Electrical, Computer & Energy Engineering Spring 4-1-2013 Low-Dimensional Signal Models

More information

MIT 9.520/6.860, Fall 2017 Statistical Learning Theory and Applications. Class 19: Data Representation by Design

MIT 9.520/6.860, Fall 2017 Statistical Learning Theory and Applications. Class 19: Data Representation by Design MIT 9.520/6.860, Fall 2017 Statistical Learning Theory and Applications Class 19: Data Representation by Design What is data representation? Let X be a data-space X M (M) F (M) X A data representation

More information

Recent Developments in Compressed Sensing

Recent Developments in Compressed Sensing Recent Developments in Compressed Sensing M. Vidyasagar Distinguished Professor, IIT Hyderabad m.vidyasagar@iith.ac.in, www.iith.ac.in/ m vidyasagar/ ISL Seminar, Stanford University, 19 April 2018 Outline

More information

Optimisation Combinatoire et Convexe.

Optimisation Combinatoire et Convexe. Optimisation Combinatoire et Convexe. Low complexity models, l 1 penalties. A. d Aspremont. M1 ENS. 1/36 Today Sparsity, low complexity models. l 1 -recovery results: three approaches. Extensions: matrix

More information

Model-Based Compressive Sensing for Signal Ensembles. Marco F. Duarte Volkan Cevher Richard G. Baraniuk

Model-Based Compressive Sensing for Signal Ensembles. Marco F. Duarte Volkan Cevher Richard G. Baraniuk Model-Based Compressive Sensing for Signal Ensembles Marco F. Duarte Volkan Cevher Richard G. Baraniuk Concise Signal Structure Sparse signal: only K out of N coordinates nonzero model: union of K-dimensional

More information

sparse and low-rank tensor recovery Cubic-Sketching

sparse and low-rank tensor recovery Cubic-Sketching Sparse and Low-Ran Tensor Recovery via Cubic-Setching Guang Cheng Department of Statistics Purdue University www.science.purdue.edu/bigdata CCAM@Purdue Math Oct. 27, 2017 Joint wor with Botao Hao and Anru

More information

Sparse Solutions of Systems of Equations and Sparse Modelling of Signals and Images

Sparse Solutions of Systems of Equations and Sparse Modelling of Signals and Images Sparse Solutions of Systems of Equations and Sparse Modelling of Signals and Images Alfredo Nava-Tudela ant@umd.edu John J. Benedetto Department of Mathematics jjb@umd.edu Abstract In this project we are

More information

Supremum of simple stochastic processes

Supremum of simple stochastic processes Subspace embeddings Daniel Hsu COMS 4772 1 Supremum of simple stochastic processes 2 Recap: JL lemma JL lemma. For any ε (0, 1/2), point set S R d of cardinality 16 ln n S = n, and k N such that k, there

More information

Compressive Sensing with Random Matrices

Compressive Sensing with Random Matrices Compressive Sensing with Random Matrices Lucas Connell University of Georgia 9 November 017 Lucas Connell (University of Georgia) Compressive Sensing with Random Matrices 9 November 017 1 / 18 Overview

More information

Massive MIMO: Signal Structure, Efficient Processing, and Open Problems II

Massive MIMO: Signal Structure, Efficient Processing, and Open Problems II Massive MIMO: Signal Structure, Efficient Processing, and Open Problems II Mahdi Barzegar Communications and Information Theory Group (CommIT) Technische Universität Berlin Heisenberg Communications and

More information

Registration-guided least-squares waveform inversion

Registration-guided least-squares waveform inversion Registration-guided least-squares waveform inversion Hyoungsu Baek 1, Henri Calandra, Laurent Demanet 1 1 MIT Mathematics department, TOTAL S.A. January 15 013 Abstract Full waveform inversion with frequency

More information

Constructing Explicit RIP Matrices and the Square-Root Bottleneck

Constructing Explicit RIP Matrices and the Square-Root Bottleneck Constructing Explicit RIP Matrices and the Square-Root Bottleneck Ryan Cinoman July 18, 2018 Ryan Cinoman Constructing Explicit RIP Matrices July 18, 2018 1 / 36 Outline 1 Introduction 2 Restricted Isometry

More information

Sparse analysis Lecture VII: Combining geometry and combinatorics, sparse matrices for sparse signal recovery

Sparse analysis Lecture VII: Combining geometry and combinatorics, sparse matrices for sparse signal recovery Sparse analysis Lecture VII: Combining geometry and combinatorics, sparse matrices for sparse signal recovery Anna C. Gilbert Department of Mathematics University of Michigan Sparse signal recovery measurements:

More information

Combining geometry and combinatorics

Combining geometry and combinatorics Combining geometry and combinatorics A unified approach to sparse signal recovery Anna C. Gilbert University of Michigan joint work with R. Berinde (MIT), P. Indyk (MIT), H. Karloff (AT&T), M. Strauss

More information

Research Project Report

Research Project Report Research Project Report Title: Prediction of pre-critical seismograms from post-critical traces Principal Investigator: Co-principal Investigators: Mrinal Sen Arthur Weglein and Paul Stoffa Final report

More information

1-Bit Compressive Sensing

1-Bit Compressive Sensing 1-Bit Compressive Sensing Petros T. Boufounos, Richard G. Baraniuk Rice University, Electrical and Computer Engineering 61 Main St. MS 38, Houston, TX 775 Abstract Compressive sensing is a new signal acquisition

More information

Convex Optimization and l 1 -minimization

Convex Optimization and l 1 -minimization Convex Optimization and l 1 -minimization Sangwoon Yun Computational Sciences Korea Institute for Advanced Study December 11, 2009 2009 NIMS Thematic Winter School Outline I. Convex Optimization II. l

More information

Simultaneous estimation of wavefields & medium parameters

Simultaneous estimation of wavefields & medium parameters Simultaneous estimation of wavefields & medium parameters reduced-space versus full-space waveform inversion Bas Peters, Felix J. Herrmann Workshop W- 2, The limit of FWI in subsurface parameter recovery.

More information

CoSaMP. Iterative signal recovery from incomplete and inaccurate samples. Joel A. Tropp

CoSaMP. Iterative signal recovery from incomplete and inaccurate samples. Joel A. Tropp CoSaMP Iterative signal recovery from incomplete and inaccurate samples Joel A. Tropp Applied & Computational Mathematics California Institute of Technology jtropp@acm.caltech.edu Joint with D. Needell

More information

Design of Projection Matrix for Compressive Sensing by Nonsmooth Optimization

Design of Projection Matrix for Compressive Sensing by Nonsmooth Optimization Design of Proection Matrix for Compressive Sensing by Nonsmooth Optimization W.-S. Lu T. Hinamoto Dept. of Electrical & Computer Engineering Graduate School of Engineering University of Victoria Hiroshima

More information

Reconstruction from Anisotropic Random Measurements

Reconstruction from Anisotropic Random Measurements Reconstruction from Anisotropic Random Measurements Mark Rudelson and Shuheng Zhou The University of Michigan, Ann Arbor Coding, Complexity, and Sparsity Workshop, 013 Ann Arbor, Michigan August 7, 013

More information