10. Multi-objective least squares
|
|
- Candice Jackson
- 6 years ago
- Views:
Transcription
1 L Vandenberghe ECE133A (Winter 2018) 10 Multi-objective least squares multi-objective least squares regularized data fitting control estimation and inversion 10-1
2 Multi-objective least squares we have several objectives J 1 = A 1 x b 1 2,, J k = A k x b k 2 A i is an m i n matrix, b i is an m i -vector we seek one x that makes all k objectives small usually there is a trade-off: no single x minimizes all objectives simultaneously Weighted least squares formulation: find x that minimizes λ 1 A 1 x b λ k A k x b k 2 coefficients λ 1,, λ k are positive weights weights λ i express relative importance of different objectives without loss of generality, we can choose λ 1 = 1 Multi-objective least squares 10-2
3 Solution of weighted least squares weighted least squares is equivalent to a standard least squares problem minimize λ1 A 1 λ2 A 2 λk A k x λ1 b 1 λ2 b 2 λk b k 2 solution is unique if the stacked matrix has linearly independent columns each matrix A i may have linearly dependent columns (or be a wide matrix) it the stacked matrix has linearly independent columns, the solution is ˆx = ( λ 1 A T 1 A λ k A T k A k ) 1 ( λ1 A T 1 b λ k A T k b k ) Multi-objective least squares 10-3
4 Example with two objectives A 1 and A 2 are 10 5 minimize A 1 x b λ A 2 x b ˆx 1 (λ) ˆx 2 (λ) ˆx 3 (λ) ˆx 4 (λ) ˆx 5 (λ) λ plot shows weighted least squares solution ˆx(λ) as function of weight λ Multi-objective least squares 10-4
5 Example with two objectives minimize A 1 x b λ A 2 x b J2(λ) λ = J 1 (λ) J 2 (λ) 6 4 λ = 1 λ = λ J 1 (λ) left figure shows J 1 (λ) = A 1ˆx(λ) b 1 2 and J 2 (λ) = A 2ˆx(λ) b 2 2 right figure shows optimal trade-off curve of J 2 (λ) versus J 1 (λ) Multi-objective least squares 10-5
6 Outline multi-objective least squares regularized data fitting control estimation and inversion
7 Motivation consider linear-in-parameters model ˆf(x) = θ 1 f 1 (x) + + θ p f p (x) we assume f 1 (x) is the constant function 1 we fit the model ˆf(x) to examples (x (1), y (1) ),, (x (N), y (N) ) large coefficient θ i makes model more sensitive to changes in f i (x) keeping θ 2,, θ p small helps avoid over-fitting this leads to two objectives: J 1 (θ) = N ( ˆf(x (k) ) y (k) ) 2, J 2 (θ) = k=1 p j=2 θ 2 j primary objective J 1 (θ) is sum of squares of prediction errors Multi-objective least squares 10-6
8 Weighted least squares formulation minimize J 1 (θ) + λj 2 (θ) = λ is positive regularization parameter N ( ˆf(x (k) ) y (k) ) 2 + λ k=1 equivalent to least squares problem: minimize [ A 1 λa2 ] θ [ y d 0 ] 2 p j=2 θ 2 j with y d = (y (1),, y (N) ), A 1 = 1 f 2 (x (1) ) f p (x (1) ) 1 f 2 (x (2) ) f p (x (2) ) 1 f 2 (x (N) ) f p (x (N) ), A 2 = stacked matrix has linearly independent columns (for positive λ) value of λ can be chosen by out-of-sample validation or cross-validation Multi-objective least squares 10-7
9 Example Train Test x solid line is signal used to generate synthetic (simulated) data 10 blue points are used as training set; 20 red points are used as test set we fit a model with five parameters θ 1,, θ 5 : ˆf(x) = θ θ k+1 cos(ω k x + φ k ) (with given ω k, φ k ) k=1 Multi-objective least squares 10-8
10 Result of regularized least squares fit Train Test RMS error versus λ 2 1 Coefficients versus λ θ 1 θ 2 θ 3 θ 4 θ λ λ minimum test RMS error is for λ around 008 increasing λ shrinks the coefficients θ 2,, θ 5 dashed lines show coefficients used to generate the data for λ near 08, estimated coefficients are close to these true values Multi-objective least squares 10-9
11 Outline multi-objective least squares regularized data fitting control estimation and inversion
12 Control y = Ax + b x is n-vector of actions or inputs y is m-vector of results or outputs relation between inputs and outputs is known affine function goal is to choose inputs to optimize different objectives on x and y Multi-objective least squares 10-10
13 Optimal input design Linear dynamical system y(t) = h 0 u(t) + h 1 u(t 1) + h 2 u(t 2) + + h t u(0) output y(t) and input u(t) are scalar we assume input u(t) is zero for t < 0 coefficients h 0, h 1, are the impulse response coefficients output is convolution of input with impulse response Optimal input design optimization variable is the input sequence x = (u(0), u(1),, u(n)) goal is to track a desired output using a small and slowly varying input Multi-objective least squares 10-11
14 Input design objectives minimize J t (x) + λ v J v (x) + λ m J m (x) track desired output y des over an interval [0, N]: J t (x) = N (y(t) y des (t)) 2 t=0 use a small and slowly varying input signal: J m (x) = N u(t) 2, J v (x) = t=0 (u(t + 1) u(t)) 2 N 1 t=0 Multi-objective least squares 10-12
15 Tracking error N J t (x) = (y(t) y des (t)) 2 t=0 = A t x b t 2 with A t = h h 1 h h 2 h 1 h h N 1 h N 2 h N 3 h 0 0 h N h N 1 h N 2 h 1 h 0, b t = y des (0) y des (1) y des (2) y des (N 1) y des (N) Multi-objective least squares 10-13
16 Input variation and magnitude Input variation J v (x) = (u(t + 1) u(t)) 2 = Dx 2 N 1 t=0 with D the N (N + 1) matrix D = Input magnitude J m (x) = N u(t) 2 = x 2 t=0 Multi-objective least squares 10-14
17 Example λ v = 0, small λ m u(t) t y(t) t larger λ v u(t) 0 2 y(t) t t Multi-objective least squares 10-15
18 Outline multi-objective least squares regularized data fitting control estimation and inversion
19 Estimation Linear measurement model y = Ax + v n-vector x contains parameters that we want to estimate m-vector v is unknown measurement error or noise m-vector y contains measurements m n matrix A relates measurements and parameters Least squares estimate: use as estimate of x the solution ˆx of minimize Ax y 2 Multi-objective least squares 10-16
20 Regularized estimation add other terms to Ax y 2 to include information about parameters Example: Tikhonov regularization minimize Ax y 2 + λ x 2 goal is to make Ax y small with small x equivalent to solving (A T A + λi)x = A T y solution is unique (if λ > 0) even when A has linearly dependent columns Multi-objective least squares 10-17
21 Signal denoising observed signal is n-vector 15 y = x + v x is unknown signal v is noise Least squares denoising yk k minimize n 1 x y 2 + λ (x i+1 x i ) 2 i=1 goal is to find slowly varying signal ˆx, close to observed signal y Multi-objective least squares 10-18
22 Matrix formulation minimize [ I λd ] x [ y 0 ] 2 D is (n 1) n finite difference matrix D = equivalent to linear equation (I + λd T D)x = y Multi-objective least squares 10-19
23 Trade-off the two objectives ˆx(λ) y and Dˆx(λ) for varying λ ˆx(λ) y Dˆx(λ) Dˆx(λ) 1 05 λ = λ λ = 10 2 λ = ˆx(λ) y Multi-objective least squares 10-20
24 Three solutions 15 λ = λ = 10 2 ˆx(λ)k 1 ˆx(λ)k k k 15 λ = 10 5 ˆx(λ) y for λ 0 ˆx(λ) avg(y)1 for λ ˆx(λ)k 1 λ 10 2 is good compromise Multi-objective least squares k
25 Image deblurring y = Ax + v x is unknown image, y is observed image A is (known) blurring matrix, v is (unknown) noise images are M N, stored as MN-vectors blurred, noisy image y deblurred image ˆx Multi-objective least squares 10-22
26 Least squares deblurring minimize Ax y 2 + λ( D v x 2 + D h x 2 ) 1st term is data fidelity term: ensures Aˆx y 2nd term penalizes differences between values at neighboring pixels D h x 2 + D v x 2 M N 1 = (X i,j+1 X ij ) 2 + i=1 j=1 M 1 i=1 N (X i+1,j X ij ) 2 j=1 if X is the M N image stored in the MN-vector x Multi-objective least squares 10-23
27 Differencing operations in matrix notation suppose x is the M N image X, stored column-wise as MN-vector x = (X 1:M,1, X 1:M,2,, X 1:M,N ) horizontal differencing: (N 1) N block matrix with M M blocks D h = I I I I I I vertical differencing: N N block matrix with (M 1) M blocks D v = D D D, D = Multi-objective least squares 10-24
28 Deblurred images Multi-objective least squares λ = 10 6 λ = 10 4 λ = 10 2 λ=
29 Tomography y = Ax + v x represents values of some quantity in a region of interest of n voxels (pixels) y represents measurements of the integral along lines through the region y i = n A ij x j + v i i=1 A ij is the length of the intersection of the line in measurement i with voxel j x 1 x 2 x 6 line for measurement i Multi-objective least squares 10-26
30 Tomographic reconstruction minimize Ax y 2 + λ( D v x 2 + D h x 2 ) D v and D h are defined as in image deblurring example on page Example left: 4000 lines (100 points, 40 lines per point) right: object placed in the square region at the center of the picture on the left region of interest is divided in pixels Multi-objective least squares 10-27
31 Regularized least squares reconstruction λ = 10 2 λ = 10 1 λ = 1 λ = 5 λ = 10 λ = 100 Multi-objective least squares 10-28
6. Approximation and fitting
6. Approximation and fitting Convex Optimization Boyd & Vandenberghe norm approximation least-norm problems regularized approximation robust approximation 6 Norm approximation minimize Ax b (A R m n with
More informationLecture 2: Tikhonov-Regularization
Lecture 2: Tikhonov-Regularization Bastian von Harrach harrach@math.uni-stuttgart.de Chair of Optimization and Inverse Problems, University of Stuttgart, Germany Advanced Instructional School on Theoretical
More informationIntroduction 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 informationLinear Inverse Problems
Linear Inverse Problems Ajinkya Kadu Utrecht University, The Netherlands February 26, 2018 Outline Introduction Least-squares Reconstruction Methods Examples Summary Introduction 2 What are inverse problems?
More informationTikhonov Regularization in General Form 8.1
Tikhonov Regularization in General Form 8.1 To introduce a more general formulation, let us return to the continuous formulation of the first-kind Fredholm integral equation. In this setting, the residual
More informationAdaptive Corrected Procedure for TVL1 Image Deblurring under Impulsive Noise
Adaptive Corrected Procedure for TVL1 Image Deblurring under Impulsive Noise Minru Bai(x T) College of Mathematics and Econometrics Hunan University Joint work with Xiongjun Zhang, Qianqian Shao June 30,
More information4. Matrix inverses. left and right inverse. linear independence. nonsingular matrices. matrices with linearly independent columns
L. Vandenberghe ECE133A (Winter 2018) 4. Matrix inverses left and right inverse linear independence nonsingular matrices matrices with linearly independent columns matrices with linearly independent rows
More information12. Cholesky factorization
L. Vandenberghe ECE133A (Winter 2018) 12. Cholesky factorization positive definite matrices examples Cholesky factorization complex positive definite matrices kernel methods 12-1 Definitions a symmetric
More information5. Orthogonal matrices
L Vandenberghe EE133A (Spring 2017) 5 Orthogonal matrices matrices with orthonormal columns orthogonal matrices tall matrices with orthonormal columns complex matrices with orthonormal columns 5-1 Orthonormal
More informationA Localized Linearized ROF Model for Surface Denoising
1 2 3 4 A Localized Linearized ROF Model for Surface Denoising Shingyu Leung August 7, 2008 5 Abstract 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 1 Introduction CT/MRI scan becomes a very
More informationImage processing and Computer Vision
1 / 1 Image processing and Computer Vision Continuous Optimization and applications to image processing Martin de La Gorce martin.de-la-gorce@enpc.fr February 2015 Optimization 2 / 1 We have a function
More informationMarkov Random Fields
Markov Random Fields Umamahesh Srinivas ipal Group Meeting February 25, 2011 Outline 1 Basic graph-theoretic concepts 2 Markov chain 3 Markov random field (MRF) 4 Gauss-Markov random field (GMRF), and
More information1. Vectors. notation. examples. vector operations. linear functions. complex vectors. complexity of vector computations
L. Vandenberghe ECE133A (Winter 2018) 1. Vectors notation examples vector operations linear functions complex vectors complexity of vector computations 1-1 Vector a vector is an ordered finite list of
More informationLecture 1 Introduction
L. Vandenberghe EE236A (Fall 2013-14) Lecture 1 Introduction course overview linear optimization examples history approximate syllabus basic definitions linear optimization in vector and matrix notation
More informationLaplace-distributed increments, the Laplace prior, and edge-preserving regularization
J. Inverse Ill-Posed Probl.? (????), 1 15 DOI 1.1515/JIIP.????.??? de Gruyter???? Laplace-distributed increments, the Laplace prior, and edge-preserving regularization Johnathan M. Bardsley Abstract. For
More information13. Nonlinear least squares
L. Vandenberghe ECE133A (Fall 2018) 13. Nonlinear least squares definition and examples derivatives and optimality condition Gauss Newton method Levenberg Marquardt method 13.1 Nonlinear least squares
More informationPhysics-based Prior modeling in Inverse Problems
Physics-based Prior modeling in Inverse Problems MURI Meeting 2013 M Usman Sadiq, Purdue University Charles A. Bouman, Purdue University In collaboration with: Jeff Simmons, AFRL Venkat Venkatakrishnan,
More informationNewton s Method for Estimating the Regularization Parameter for Least Squares: Using the Chi-curve
Newton s Method for Estimating the Regularization Parameter for Least Squares: Using the Chi-curve Rosemary Renaut, Jodi Mead Arizona State and Boise State Copper Mountain Conference on Iterative Methods
More information9. Least squares data fitting
L. Vandenberghe EE133A (Spring 2017) 9. Least squares data fitting model fitting regression linear-in-parameters models time series examples validation least squares classification statistics interpretation
More informationFinal Exam. Problem Score Problem Score 1 /10 8 /10 2 /10 9 /10 3 /10 10 /10 4 /10 11 /10 5 /10 12 /10 6 /10 13 /10 7 /10 Total /130
EE103/CME103: Introduction to Matrix Methods December 9 2015 S. Boyd Final Exam You may not use any books, notes, or computer programs (e.g., Julia). Throughout this exam we use standard mathematical notation;
More informationLecture 5 Least-squares
EE263 Autumn 2008-09 Stephen Boyd Lecture 5 Least-squares least-squares (approximate) solution of overdetermined equations projection and orthogonality principle least-squares estimation BLUE property
More informationInverse problem and optimization
Inverse problem and optimization Laurent Condat, Nelly Pustelnik CNRS, Gipsa-lab CNRS, Laboratoire de Physique de l ENS de Lyon Decembre, 15th 2016 Inverse problem and optimization 2/36 Plan 1. Examples
More informationComputer Vision & Digital Image Processing
Computer Vision & Digital Image Processing Image Restoration and Reconstruction I Dr. D. J. Jackson Lecture 11-1 Image restoration Restoration is an objective process that attempts to recover an image
More informationRegularization methods for large-scale, ill-posed, linear, discrete, inverse problems
Regularization methods for large-scale, ill-posed, linear, discrete, inverse problems Silvia Gazzola Dipartimento di Matematica - Università di Padova January 10, 2012 Seminario ex-studenti 2 Silvia Gazzola
More informationLeast Squares. Stephen Boyd. EE103 Stanford University. October 28, 2017
Least Squares Stephen Boyd EE103 Stanford University October 28, 2017 Outline Least squares problem Solution of least squares problem Examples Least squares problem 2 Least squares problem suppose m n
More informationLinear Operators and Fourier Transform
Linear Operators and Fourier Transform DD2423 Image Analysis and Computer Vision Mårten Björkman Computational Vision and Active Perception School of Computer Science and Communication November 13, 2013
More informationImage processing and nonparametric regression
Image processing and nonparametric regression Rencontres R BoRdeaux 2012 B. Thieurmel Collaborators : P.A. Cornillon, N. Hengartner, E. Matzner-Løber, B. Wolhberg 2 Juillet 2012 Rencontres R BoRdeaux 2012
More informationHomework 6 Solutions
8-290 Signals and Systems Profs. Byron Yu and Pulkit Grover Fall 208 Homework 6 Solutions. Part One. (2 points) Consider an LTI system with impulse response h(t) e αt u(t), (a) Compute the frequency response
More informationA Tutorial on Recursive methods in Linear Least Squares Problems
A Tutorial on Recursive methods in Linear Least Squares Problems by Arvind Yedla 1 Introduction This tutorial motivates the use of Recursive Methods in Linear Least Squares problems, specifically Recursive
More information14. Nonlinear equations
L. Vandenberghe ECE133A (Winter 2018) 14. Nonlinear equations Newton method for nonlinear equations damped Newton method for unconstrained minimization Newton method for nonlinear least squares 14-1 Set
More informationPDEs in Image Processing, Tutorials
PDEs in Image Processing, Tutorials Markus Grasmair Vienna, Winter Term 2010 2011 Direct Methods Let X be a topological space and R: X R {+ } some functional. following definitions: The mapping R is lower
More informationConvex 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 informationRegularization Parameter Estimation for Least Squares: A Newton method using the χ 2 -distribution
Regularization Parameter Estimation for Least Squares: A Newton method using the χ 2 -distribution Rosemary Renaut, Jodi Mead Arizona State and Boise State September 2007 Renaut and Mead (ASU/Boise) Scalar
More informationCS 179: LECTURE 16 MODEL COMPLEXITY, REGULARIZATION, AND CONVOLUTIONAL NETS
CS 179: LECTURE 16 MODEL COMPLEXITY, REGULARIZATION, AND CONVOLUTIONAL NETS LAST TIME Intro to cudnn Deep neural nets using cublas and cudnn TODAY Building a better model for image classification Overfitting
More informationInverse Ill Posed Problems in Image Processing
Inverse Ill Posed Problems in Image Processing Image Deblurring I. Hnětynková 1,M.Plešinger 2,Z.Strakoš 3 hnetynko@karlin.mff.cuni.cz, martin.plesinger@tul.cz, strakos@cs.cas.cz 1,3 Faculty of Mathematics
More informationLeast Squares Data Fitting
Least Squares Data Fitting Stephen Boyd EE103 Stanford University October 31, 2017 Outline Least squares model fitting Validation Feature engineering Least squares model fitting 2 Setup we believe a scalar
More informationNonlinear Least Squares
Nonlinear Least Squares Stephen Boyd EE103 Stanford University December 6, 2016 Outline Nonlinear equations and least squares Examples Levenberg-Marquardt algorithm Nonlinear least squares classification
More informationGEORGIA INSTITUTE OF TECHNOLOGY SCHOOL of ELECTRICAL & COMPUTER ENGINEERING FINAL EXAM. COURSE: ECE 3084A (Prof. Michaels)
GEORGIA INSTITUTE OF TECHNOLOGY SCHOOL of ELECTRICAL & COMPUTER ENGINEERING FINAL EXAM DATE: 30-Apr-14 COURSE: ECE 3084A (Prof. Michaels) NAME: STUDENT #: LAST, FIRST Write your name on the front page
More informationNovel integro-differential equations in image processing and its applications
Novel integro-differential equations in image processing and its applications Prashant Athavale a and Eitan Tadmor b a Institute of Pure and Applied Mathematics, University of California, Los Angeles,
More informationMathematical Beer Goggles or The Mathematics of Image Processing
How Mathematical Beer Goggles or The Mathematics of Image Processing Department of Mathematical Sciences University of Bath Postgraduate Seminar Series University of Bath 12th February 2008 1 How 2 How
More informationThe Chi-squared Distribution of the Regularized Least Squares Functional for Regularization Parameter Estimation
The Chi-squared Distribution of the Regularized Least Squares Functional for Regularization Parameter Estimation Rosemary Renaut Collaborators: Jodi Mead and Iveta Hnetynkova DEPARTMENT OF MATHEMATICS
More informationUncertainty Models in Quasiconvex Optimization for Geometric Reconstruction
Uncertainty Models in Quasiconvex Optimization for Geometric Reconstruction Qifa Ke and Takeo Kanade Department of Computer Science, Carnegie Mellon University Email: ke@cmu.edu, tk@cs.cmu.edu Abstract
More informationErkut Erdem. Hacettepe University February 24 th, Linear Diffusion 1. 2 Appendix - The Calculus of Variations 5.
LINEAR DIFFUSION Erkut Erdem Hacettepe University February 24 th, 2012 CONTENTS 1 Linear Diffusion 1 2 Appendix - The Calculus of Variations 5 References 6 1 LINEAR DIFFUSION The linear diffusion (heat)
More informationStatistically-Based Regularization Parameter Estimation for Large Scale Problems
Statistically-Based Regularization Parameter Estimation for Large Scale Problems Rosemary Renaut Joint work with Jodi Mead and Iveta Hnetynkova March 1, 2010 National Science Foundation: Division of Computational
More informationFourier Transforms 1D
Fourier Transforms 1D 3D Image Processing Alireza Ghane 1 Overview Recap Intuitions Function representations shift-invariant spaces linear, time-invariant (LTI) systems complex numbers Fourier Transforms
More informationFinite Volume Method for Scalar Advection in 2D
Chapter 9 Finite Volume Method for Scalar Advection in D 9. Introduction The purpose of this exercise is to code a program to integrate the scalar advection equation in two-dimensional flows. 9. The D
More informationA memory gradient algorithm for l 2 -l 0 regularization with applications to image restoration
A memory gradient algorithm for l 2 -l 0 regularization with applications to image restoration E. Chouzenoux, A. Jezierska, J.-C. Pesquet and H. Talbot Université Paris-Est Lab. d Informatique Gaspard
More informationIMAGE ENHANCEMENT II (CONVOLUTION)
MOTIVATION Recorded images often exhibit problems such as: blurry noisy Image enhancement aims to improve visual quality Cosmetic processing Usually empirical techniques, with ad hoc parameters ( whatever
More informationAdvanced Numerical Linear Algebra: Inverse Problems
Advanced Numerical Linear Algebra: Inverse Problems Rosemary Renaut Spring 23 Some Background on Inverse Problems Constructing PSF Matrices The DFT Rosemary Renaut February 4, 23 References Deblurring
More informationThe Chi-squared Distribution of the Regularized Least Squares Functional for Regularization Parameter Estimation
The Chi-squared Distribution of the Regularized Least Squares Functional for Regularization Parameter Estimation Rosemary Renaut DEPARTMENT OF MATHEMATICS AND STATISTICS Prague 2008 MATHEMATICS AND STATISTICS
More informationOptimisation in imaging
Optimisation in imaging Hugues Talbot ICIP 2014 Tutorial Optimisation on Hierarchies H. Talbot : Optimisation 1/59 Outline of the lecture 1 Concepts in optimization Cost function Constraints Duality 2
More informationIterative Methods for Ill-Posed Problems
Iterative Methods for Ill-Posed Problems Based on joint work with: Serena Morigi Fiorella Sgallari Andriy Shyshkov Salt Lake City, May, 2007 Outline: Inverse and ill-posed problems Tikhonov regularization
More informationTwo well known examples. Applications of structured low-rank approximation. Approximate realisation = Model reduction. System realisation
Two well known examples Applications of structured low-rank approximation Ivan Markovsky System realisation Discrete deconvolution School of Electronics and Computer Science University of Southampton The
More information2 Regularized Image Reconstruction for Compressive Imaging and Beyond
EE 367 / CS 448I Computational Imaging and Display Notes: Compressive Imaging and Regularized Image Reconstruction (lecture ) Gordon Wetzstein gordon.wetzstein@stanford.edu This document serves as a supplement
More informationNon-negative Quadratic Programming Total Variation Regularization for Poisson Vector-Valued Image Restoration
University of New Mexico UNM Digital Repository Electrical & Computer Engineering Technical Reports Engineering Publications 5-10-2011 Non-negative Quadratic Programming Total Variation Regularization
More informationOverview of Spatial Statistics with Applications to fmri
with Applications to fmri School of Mathematics & Statistics Newcastle University April 8 th, 2016 Outline Why spatial statistics? Basic results Nonstationary models Inference for large data sets An example
More informationInvestigating the Influence of Box-Constraints on the Solution of a Total Variation Model via an Efficient Primal-Dual Method
Article Investigating the Influence of Box-Constraints on the Solution of a Total Variation Model via an Efficient Primal-Dual Method Andreas Langer Department of Mathematics, University of Stuttgart,
More informationVector spaces. DS-GA 1013 / MATH-GA 2824 Optimization-based Data Analysis.
Vector spaces DS-GA 1013 / MATH-GA 2824 Optimization-based Data Analysis http://www.cims.nyu.edu/~cfgranda/pages/obda_fall17/index.html Carlos Fernandez-Granda Vector space Consists of: A set V A scalar
More informationDiscrete-time Fourier transform (DTFT) representation of DT aperiodic signals Section The (DT) Fourier transform (or spectrum) of x[n] is
Discrete-time Fourier transform (DTFT) representation of DT aperiodic signals Section 5. 3 The (DT) Fourier transform (or spectrum) of x[n] is X ( e jω) = n= x[n]e jωn x[n] can be reconstructed from its
More informationGolub-Kahan iterative bidiagonalization and determining the noise level in the data
Golub-Kahan iterative bidiagonalization and determining the noise level in the data Iveta Hnětynková,, Martin Plešinger,, Zdeněk Strakoš, * Charles University, Prague ** Academy of Sciences of the Czech
More informationCS540 Machine learning Lecture 5
CS540 Machine learning Lecture 5 1 Last time Basis functions for linear regression Normal equations QR SVD - briefly 2 This time Geometry of least squares (again) SVD more slowly LMS Ridge regression 3
More informationLecture 6. Numerical methods. Approximation of functions
Lecture 6 Numerical methods Approximation of functions Lecture 6 OUTLINE 1. Approximation and interpolation 2. Least-square method basis functions design matrix residual weighted least squares normal equation
More informationLPA-ICI Applications in Image Processing
LPA-ICI Applications in Image Processing Denoising Deblurring Derivative estimation Edge detection Inverse halftoning Denoising Consider z (x) =y (x)+η (x), wherey is noise-free image and η is noise. assume
More informationTime Series. Karanveer Mohan Keegan Go Stephen Boyd. EE103 Stanford University. November 15, 2016
Time Series Karanveer Mohan Keegan Go Stephen Boyd EE103 Stanford University November 15, 2016 Outline Introduction Linear operations Least-squares Prediction Introduction 2 Time series data represent
More informationLearning Deep Architectures for AI. Part II - Vijay Chakilam
Learning Deep Architectures for AI - Yoshua Bengio Part II - Vijay Chakilam Limitations of Perceptron x1 W, b 0,1 1,1 y x2 weight plane output =1 output =0 There is no value for W and b such that the model
More informationIll Posed Inverse Problems in Image Processing
Ill Posed Inverse Problems in Image Processing Introduction, Structured matrices, Spectral filtering, Regularization, Noise revealing I. Hnětynková 1,M.Plešinger 2,Z.Strakoš 3 hnetynko@karlin.mff.cuni.cz,
More informationNONLINEAR DIFFUSION PDES
NONLINEAR DIFFUSION PDES Erkut Erdem Hacettepe University March 5 th, 0 CONTENTS Perona-Malik Type Nonlinear Diffusion Edge Enhancing Diffusion 5 References 7 PERONA-MALIK TYPE NONLINEAR DIFFUSION The
More informationNewton s method for obtaining the Regularization Parameter for Regularized Least Squares
Newton s method for obtaining the Regularization Parameter for Regularized Least Squares Rosemary Renaut, Jodi Mead Arizona State and Boise State International Linear Algebra Society, Cancun Renaut and
More informationANALYTICAL BOUNDS ON THE MINIMIZERS OF (NONCONVEX) REGULARIZED LEAST-SQUARES. Mila Nikolova. (Communicated by Jackie (Jianhong) Shen)
Inverse Problems and Imaging Volume 1, No. 4, 2007, 661 677 Web site: http://www.aimsciences.org ANALYTICAL BOUNDS ON THE MINIMIZERS OF NONCONVEX) REGULARIZED LEAST-SQUARES Mila Nikolova CMLA, ENS Cachan,
More informationLeast-squares data fitting
EE263 Autumn 2015 S. Boyd and S. Lall Least-squares data fitting 1 Least-squares data fitting we are given: functions f 1,..., f n : S R, called regressors or basis functions data or measurements (s i,
More informationData Sparse Matrix Computation - Lecture 20
Data Sparse Matrix Computation - Lecture 20 Yao Cheng, Dongping Qi, Tianyi Shi November 9, 207 Contents Introduction 2 Theorems on Sparsity 2. Example: A = [Φ Ψ]......................... 2.2 General Matrix
More informationNAME: ht () 1 2π. Hj0 ( ) dω Find the value of BW for the system having the following impulse response.
University of California at Berkeley Department of Electrical Engineering and Computer Sciences Professor J. M. Kahn, EECS 120, Fall 1998 Final Examination, Wednesday, December 16, 1998, 5-8 pm NAME: 1.
More informationLearning Spectral Graph Segmentation
Learning Spectral Graph Segmentation AISTATS 2005 Timothée Cour Jianbo Shi Nicolas Gogin Computer and Information Science Department University of Pennsylvania Computer Science Ecole Polytechnique Graph-based
More informationControl Design. Lecture 9: State Feedback and Observers. Two Classes of Control Problems. State Feedback: Problem Formulation
Lecture 9: State Feedback and s [IFAC PB Ch 9] State Feedback s Disturbance Estimation & Integral Action Control Design Many factors to consider, for example: Attenuation of load disturbances Reduction
More informationConvex Optimization: Applications
Convex Optimization: Applications Lecturer: Pradeep Ravikumar Co-instructor: Aarti Singh Convex Optimization 1-75/36-75 Based on material from Boyd, Vandenberghe Norm Approximation minimize Ax b (A R m
More informationLaplacian Filters. Sobel Filters. Laplacian Filters. Laplacian Filters. Laplacian Filters. Laplacian Filters
Sobel Filters Note that smoothing the image before applying a Sobel filter typically gives better results. Even thresholding the Sobel filtered image cannot usually create precise, i.e., -pixel wide, edges.
More informationNon-uniformity Correction of IR Images using Natural Scene Statistics
Non-uniformity Correction of IR Images using Natural Scene Statistics Todd Goodall, Alan C. Bovik, and Haris Vikalo 1 Nicholas G. Paulter, Jr. 2 1 University of Texas at Austin 2 National Institute of
More informationSparse and Robust Optimization and Applications
Sparse and and Statistical Learning Workshop Les Houches, 2013 Robust Laurent El Ghaoui with Mert Pilanci, Anh Pham EECS Dept., UC Berkeley January 7, 2013 1 / 36 Outline Sparse Sparse Sparse Probability
More informationSpatially adaptive alpha-rooting in BM3D sharpening
Spatially adaptive alpha-rooting in BM3D sharpening Markku Mäkitalo and Alessandro Foi Department of Signal Processing, Tampere University of Technology, P.O. Box FIN-553, 33101, Tampere, Finland e-mail:
More informationLeast squares problems Linear Algebra with Computer Science Application
Linear Algebra with Computer Science Application April 8, 018 1 Least Squares Problems 11 Least Squares Problems What do you do when Ax = b has no solution? Inconsistent systems arise often in applications
More informationImage Noise: Detection, Measurement and Removal Techniques. Zhifei Zhang
Image Noise: Detection, Measurement and Removal Techniques Zhifei Zhang Outline Noise measurement Filter-based Block-based Wavelet-based Noise removal Spatial domain Transform domain Non-local methods
More informationMaximal Entropy for Reconstruction of Back Projection Images
Maximal Entropy for Reconstruction of Back Projection Images Tryphon Georgiou Department of Electrical and Computer Engineering University of Minnesota Minneapolis, MN 5545 Peter J Olver Department of
More informationTowards Tomographic Photoelasticity
Towards Tomographic Photoelasticity Dr Rachel Tomlinson Department of Mechanical Engineering, Outline What is photoelasticity? 3D Photoelasticity Methods Advances in data collection and processing Future
More informationLinear Neural Networks
Chapter 10 Linear Neural Networks In this chapter, we introduce the concept of the linear neural network. 10.1 Introduction and Notation 1. The linear neural cell, or node has the schematic form as shown
More information2A1H Time-Frequency Analysis II Bugs/queries to HT 2011 For hints and answers visit dwm/courses/2tf
Time-Frequency Analysis II (HT 20) 2AH 2AH Time-Frequency Analysis II Bugs/queries to david.murray@eng.ox.ac.uk HT 20 For hints and answers visit www.robots.ox.ac.uk/ dwm/courses/2tf David Murray. A periodic
More informationPreconditioning. Noisy, Ill-Conditioned Linear Systems
Preconditioning Noisy, Ill-Conditioned Linear Systems James G. Nagy Emory University Atlanta, GA Outline 1. The Basic Problem 2. Regularization / Iterative Methods 3. Preconditioning 4. Example: Image
More informationMIT 2.71/2.710 Optics 10/31/05 wk9-a-1. The spatial frequency domain
10/31/05 wk9-a-1 The spatial frequency domain Recall: plane wave propagation x path delay increases linearly with x λ z=0 θ E 0 x exp i2π sinθ + λ z i2π cosθ λ z plane of observation 10/31/05 wk9-a-2 Spatial
More informationEE 367 / CS 448I Computational Imaging and Display Notes: Image Deconvolution (lecture 6)
EE 367 / CS 448I Computational Imaging and Display Notes: Image Deconvolution (lecture 6) Gordon Wetzstein gordon.wetzstein@stanford.edu This document serves as a supplement to the material discussed in
More informationTowards Improved Sensitivity in Feature Extraction from Signals: one and two dimensional
Towards Improved Sensitivity in Feature Extraction from Signals: one and two dimensional Rosemary Renaut, Hongbin Guo, Jodi Mead and Wolfgang Stefan Supported by Arizona Alzheimer s Research Center and
More informationWavelets and Multiresolution Processing
Wavelets and Multiresolution Processing Wavelets Fourier transform has it basis functions in sinusoids Wavelets based on small waves of varying frequency and limited duration In addition to frequency,
More informationEdge Detection in Computer Vision Systems
1 CS332 Visual Processing in Computer and Biological Vision Systems Edge Detection in Computer Vision Systems This handout summarizes much of the material on the detection and description of intensity
More informationCovariance Matrix Simplification For Efficient Uncertainty Management
PASEO MaxEnt 2007 Covariance Matrix Simplification For Efficient Uncertainty Management André Jalobeanu, Jorge A. Gutiérrez PASEO Research Group LSIIT (CNRS/ Univ. Strasbourg) - Illkirch, France *part
More informationA Riemannian Framework for Denoising Diffusion Tensor Images
A Riemannian Framework for Denoising Diffusion Tensor Images Manasi Datar No Institute Given Abstract. Diffusion Tensor Imaging (DTI) is a relatively new imaging modality that has been extensively used
More informationLecture 3 January 23
EE 123: Digital Signal Processing Spring 2007 Lecture 3 January 23 Lecturer: Prof. Anant Sahai Scribe: Dominic Antonelli 3.1 Outline These notes cover the following topics: Eigenvectors and Eigenvalues
More informationIntro. Computer Control Systems: F9
Intro. Computer Control Systems: F9 State-feedback control and observers Dave Zachariah Dept. Information Technology, Div. Systems and Control 1 / 21 dave.zachariah@it.uu.se F8: Quiz! 2 / 21 dave.zachariah@it.uu.se
More informationLecture 7 Monotonicity. September 21, 2008
Lecture 7 Monotonicity September 21, 2008 Outline Introduce several monotonicity properties of vector functions Are satisfied immediately by gradient maps of convex functions In a sense, role of monotonicity
More informationVECTORS IN A STRAIGHT LINE
A. The Equation of a Straight Line VECTORS P3 VECTORS IN A STRAIGHT LINE A particular line is uniquely located in space if : I. It has a known direction, d, and passed through a known fixed point, or II.
More informationA Brief Outline of Math 355
A Brief Outline of Math 355 Lecture 1 The geometry of linear equations; elimination with matrices A system of m linear equations with n unknowns can be thought of geometrically as m hyperplanes intersecting
More informationChapter 1 Fundamental Concepts
Chapter 1 Fundamental Concepts 1 Signals A signal is a pattern of variation of a physical quantity, often as a function of time (but also space, distance, position, etc). These quantities are usually the
More informationENERGY METHODS IN IMAGE PROCESSING WITH EDGE ENHANCEMENT
ENERGY METHODS IN IMAGE PROCESSING WITH EDGE ENHANCEMENT PRASHANT ATHAVALE Abstract. Digital images are can be realized as L 2 (R 2 objects. Noise is introduced in a digital image due to various reasons.
More information