Appendix A: An Alternative Estimation Procedure Dual Penalized Expansion

Size: px
Start display at page:

Download "Appendix A: An Alternative Estimation Procedure Dual Penalized Expansion"

Transcription

1 Supplemental Materials for Functional Linear Models for Zero-Inflated Count Data with Application to Modeling Hospitalizations in Patients on Dialysis by Şentürk, D., Dalrymple, L. S. and Nguyen, D. V. Appendix A: An Alternative Estimation Procedure Dual Penalized Expansion We also explore an alternative estimation procedure called dual penalized expansion (DPE), inspired by the work of Goldsmith et al. [20], for generalized functional linear, hurdle and ZIP models, where penalized dual (separate) basis expansions for the predictor process and the coefficient functions of interest are considered. In contrast to PR, the dual penalized expansion (DPE) approach expands the original longitudinal predictor directly on its functional principal component basis, without reconstruction of X(t). In addition, similar to PR, the regression function, β(t), is expanded on spline basis functions. Step 1: Dimension Reduction via Basis Expansion The DPE approach considers dual expansions of the predictor process, X(t), and a coefficient function, β(t), using two separate sets of basis functions: K x X i (t) µ X (t) + ξ iv ψ v (t), v=1 K b β(t) b u φ u (t), u=1 where ξ iv = {X i (t) µ X (t)}ψ v (t)dt. The basis functions ψ v (t) and φ u (t) are taken to be the functional principal components (PCs) basis and truncated power spline basis, respectively. K x and K b are chosen large and they satisfy the identifiability constraint K x K b. We take K x = K b = 10, which provides adequate approximation as was found in Goldsmith et al. [20] for sparse designs. The expansion of γ(t) in a functional hurdle or a ZIP model follows similarly to the above expansion for β(t). More precisely, γ(t) K a ϑ=1 a ϑθ ϑ (t), where θ ϑ (t) is truncated power spline basis. Using the double expansion above, K b [ Kx X i (t)β(t)dt b u {ξ iv u=1 v=1 } ψ v (t)φ u (t)dt + 1 ] µ X (t)φ u (t)dt = (ξi T J ψφ + µ)b

2 where b = (b 1,..., b Kb ) T, ξ i = (ξ i1,..., ξ ikx ) T, J ψφ is a K x K b matrix with (v, u)th entry equal to ψ v (t)φ u (t)dt, and µ is a 1 K b vector with the uth entry equal to µ X (t)φ u (t)dt. When ψ(t) and φ(t) are taken to be the same set of orthonormal basis functions, such as the truncated power series spline basis defined above for t [0, 1], J ψφ reduces to an identity matrix. For sparse longitudinal data, direct expansion of X i (t) on a spline basis is not feasible since ξ iv = {X i (t) µ X (t)}ψ v (t)dt cannot be well approximated. Thus, X i (t) is expanded by functional PCs basis functions, an approach that is suitable for sparse longitudinal data. Hence, estimation of the functional PCs basis functions, eigenscores and the mean function of X(t), as described in Section 3.1 Step 0, allows estimation of the terms in (ξi T J ψφ + µ). Similar expansions lead to X i (t)γ(t)dt (ξi T J ψφ + µ)a for the functional hurdle and ZIP models. Thus, substituting the above derived equality for X i (t)β(t)dt from above into the generalized functional linear model (1) gives g(µ i ) = β 0 + X i (t)β(t)dt + α r Z ri β 0 + (ξi T J ψφ + µ)b + α r Z ri. The reduced generalized linear model has a n (1+K b +p) design matrix, (1 n, ξj ψφ + µ, Z) and parameter vector (β 0, b, α) T, where b = (b 1,..., b Kb ) T, α = (α 1,..., α p ) T, 1 n is a n 1 vector of ones, n K x matrix ξ = (ξ 1,..., ξ n ) T and Z is the n p matrix equal to Z = (Z 1,..., Z p ) for Z r = (Z r1,..., Z rn ) T. The functional hurdle model in (3) reduces to a classical hurdle model g 1 (p i ) β 0 + (ξi T J ψφ + µ)b + g 2 (λ i ) γ 0 + (ξi T J ψφ + µ)a + α r Z ri, ζ r Z ri, and with analogously defined design matrices and parameter vectors, and with design matrix (1 n, ξj ψφ + µ, Z) and parameter vector (γ 0, a, ζ) T where a = (a 1,..., a Ka ) T, ζ = (ζ 1,..., ζ p ) T. See Section 3.1 for details. Step 2: Penalized Maximum Likelihood 2

3 The penalized maximum likelihood estimation procedure for DPE proceeds as proposed for the PR approach, described in details in step 2 of Section 3.1, with the estimated design matrix Ŵ and vector Ŵi replaced by (ˆξĴψφ + ˆ µ) and (ˆξ T i Ĵ ψφ + ˆ µ), respectively, in the linear predictor and likelihood equations. Difference between PR and DPE PR and DPE both use FPCA to expand the functional predictor and the penalized truncated power series for the regression coefficient. However, since PR uses the FPCA decomposition in the reconstruction step, it effectively uses a smaller number of functional principal components (e.g. rarely selects more than 3 components in applications) compared to the DPE expansion. The main reason for DPE using a larger number of PCs in the FPCA expansion of the predictor process is the identifiability condition involved in the dual expansion procedure. In order for the spline coefficients to be identifiable in the expansion of the regression function, the FPCA decomposition of the predictor process needs to contain at least the same number of principal components (i.e. K x K b ). We take K x = K b = 10, which provides adequate approximation as was found in Goldsmith et al. [20] for sparse designs. The number of PCs used is the main difference between PR and DPE. In the following sections we study the finite sample properties of DPE in comparison to PR and PCR via simulations. Appendix B: Simulation Results with DPE Finite sample properties of DPE were evaluated in the same simulation settings as described in Section 4.1. Independent preliminary simulation studies were carried out to select the optimal regularization parameters for DPE. Median δ values minimizing CV error across multiple runs for the generalized functional linear model fits are.0055 for both sample sizes in the sparse design and they are.0005 and.001 for the denser design at n = 200 and n = 400, respectively. The selected (δ 1, δ 2 ) regularization pairs for the functional hurdle model for the 3

4 sparse design are (.01,.01) and (.05,.03); they are (.01,.05) and (.01,.01) for the denser design at n = 200 and n = 400, respectively. For the functional ZIP model, they are (.05,.05) and (.075,.05) in the sparse and denser designs for n = 200 and are both (.05,.05) for n = 400. For convenience of comparisons, we present the Tables 1, 2 and 3 from Section 4.2 of the manuscript once more, but this time with DPE results included along with PR and PCR results. Overall, both PR and DPE perform similarly and lead to significant efficiency gains in regression function estimation in generalized functional linear and functional linear mixture models over the more standard approach of PCR. More specifically, PR and DPE perform similarly in all model parameters and simulation set-ups for the generalized functional linear model. For the regression function estimation, PR trends towards some modest efficiency gains over DPE in the functional hurdle model for both binary and zero-truncated Poisson parts for the sparse design setting at n = 200. For the regression function estimation in the functional ZIP model, PR leads to efficiency gains in the Poisson part for sparse designs for both sample sizes. 4

5 Table 1: Simulation results for generalized functional linear model. Median, mean and squared deviation error (ME) reported for all model parameters for different estimation techniques in fitting a generalized functional linear model over different sparsity levels of the longitudinal predictor and at different sample sizes in 200 Monte Carlo runs. PR: penalized reconstruction; DPE: dual penalized expansion; PCR: principal components regression. Design n Median 25% 75% Median 25% 75% Median 25% 75% ME β ME β0 ME α PR DPE PCR PR DPE PCR PR DPE PCR PR DPE PCR

6 Table 2: Simulation results for functional hurdle model. Median, mean, squared deviation error (ME) reported for all model parameters for different estimation techniques in fitting a functional hurdle model over different sparsity levels of the longitudinal predictor and at different sample sizes in 200 Monte Carlo runs. PR: penalized reconstruction; DPE: dual penalized expansion; PCR: principal components regression. Design n Median 25% 75% Median 25% 75% Median 25% 75% ME β ME β0 ME α Binary PR DPE PCR PR DPE PCR PR DPE PCR PR DPE PCR ME γ ME γ0 ME ζ Zero-truncated Poisson PR < DPE < PCR PR < DPE < PCR < PR < DPE < PCR < PR < DPE < PCR <

7 Table 3: Simulation results for functional ZIP model. Median, mean, squared deviation error (ME) reported for all model parameters for different estimation techniques in fitting a functional ZIP model over different sparsity levels of the longitudinal predictor and at different sample sizes in 200 Monte Carlo runs. PR: penalized reconstruction; DPE: dual penalized expansion; PCR: principal components regression. Design n Median 25% 75% Median 25% 75% Median 25% 75% ME β ME β0 ME α Binary PR DPE PCR PR DPE PCR PR DPE PCR PR DPE PCR ME γ ME γ0 ME ζ Poisson PR < DPE < PCR PR < DPE < PCR PR < DPE < PCR PR < <.001 < DPE <.001 < PCR

Chapter 5: Basis Expansion and Regularization

Chapter 5: Basis Expansion and Regularization Chapter 5: Basis Expansion and Regularization DD3364 April 1, 2012 Introduction Main idea Moving beyond linearity Augment the vector of inputs X with additional variables. These are transformations of

More information

Lecture 7: Splines and Generalized Additive Models

Lecture 7: Splines and Generalized Additive Models Lecture 7: and Generalized Additive Models Computational Statistics Thierry Denœux April, 2016 Introduction Overview Introduction Simple approaches Polynomials Step functions Regression splines Natural

More information

Quick Start with CASSY Lab. Bi-05-05

Quick Start with CASSY Lab. Bi-05-05 Quick Start with CASSY Lab Bi-05-05 About this manual This manual helps you getting started with the CASSY system. The manual does provide you the information you need to start quickly a simple CASSY experiment

More information

Estimation and Inference by the Method of Projection Minimum Distance. Òscar Jordà Sharon Kozicki U.C. Davis Bank of Canada

Estimation and Inference by the Method of Projection Minimum Distance. Òscar Jordà Sharon Kozicki U.C. Davis Bank of Canada Estimation and Inference by the Method of Projection Minimum Distance Òscar Jordà Sharon Kozicki U.C. Davis Bank of Canada The Paper in a Nutshell: An Efficient Limited Information Method Step 1: estimate

More information

This is called a linear basis expansion, and h m is the mth basis function For example if X is one-dimensional: f (X) = β 0 + β 1 X + β 2 X 2, or

This is called a linear basis expansion, and h m is the mth basis function For example if X is one-dimensional: f (X) = β 0 + β 1 X + β 2 X 2, or STA 450/4000 S: February 2 2005 Flexible modelling using basis expansions (Chapter 5) Linear regression: y = Xβ + ɛ, ɛ (0, σ 2 ) Smooth regression: y = f (X) + ɛ: f (X) = E(Y X) to be specified Flexible

More information

SVMs for Structured Output. Andrea Vedaldi

SVMs for Structured Output. Andrea Vedaldi SVMs for Structured Output Andrea Vedaldi SVM struct Tsochantaridis Hofmann Joachims Altun 04 Extending SVMs 3 Extending SVMs SVM = parametric function arbitrary input binary output 3 Extending SVMs SVM

More information

Outline Introduction Problem Formulation Proposed Solution Applications Conclusion. Compressed Sensing. David L Donoho Presented by: Nitesh Shroff

Outline Introduction Problem Formulation Proposed Solution Applications Conclusion. Compressed Sensing. David L Donoho Presented by: Nitesh Shroff Compressed Sensing David L Donoho Presented by: Nitesh Shroff University of Maryland Outline 1 Introduction Compressed Sensing 2 Problem Formulation Sparse Signal Problem Statement 3 Proposed Solution

More information

Simultaneous Perturbation Stochastic Approximation Algorithm Combined with Neural Network and Fuzzy Simulation

Simultaneous Perturbation Stochastic Approximation Algorithm Combined with Neural Network and Fuzzy Simulation .--- Simultaneous Perturbation Stochastic Approximation Algorithm Combined with Neural Networ and Fuzzy Simulation Abstract - - - - Keywords: Many optimization problems contain fuzzy information. Possibility

More information

Collaborative Sparsity and Compressive MRI

Collaborative Sparsity and Compressive MRI Modeling and Computation Seminar February 14, 2013 Table of Contents 1 T2 Estimation 2 Undersampling in MRI 3 Compressed Sensing 4 Model-Based Approach 5 From L1 to L0 6 Spatially Adaptive Sparsity MRI

More information

Highly Symmetric Bi-frames for Triangle Surface Multiresolution Processing

Highly Symmetric Bi-frames for Triangle Surface Multiresolution Processing Highly Symmetric Bi-frames for Triangle Surface Multiresolution Processing Qingtang Jiang and Dale K. Pounds Abstract In this paper we investigate the construction of dyadic affine (wavelet) bi-frames

More information

A fast algorithm for sparse reconstruction based on shrinkage, subspace optimization and continuation [Wen,Yin,Goldfarb,Zhang 2009]

A fast algorithm for sparse reconstruction based on shrinkage, subspace optimization and continuation [Wen,Yin,Goldfarb,Zhang 2009] A fast algorithm for sparse reconstruction based on shrinkage, subspace optimization and continuation [Wen,Yin,Goldfarb,Zhang 2009] Yongjia Song University of Wisconsin-Madison April 22, 2010 Yongjia Song

More information

Interpolation by Spline Functions

Interpolation by Spline Functions Interpolation by Spline Functions Com S 477/577 Sep 0 007 High-degree polynomials tend to have large oscillations which are not the characteristics of the original data. To yield smooth interpolating curves

More information

Software Documentation of the Potential Support Vector Machine

Software Documentation of the Potential Support Vector Machine Software Documentation of the Potential Support Vector Machine Tilman Knebel and Sepp Hochreiter Department of Electrical Engineering and Computer Science Technische Universität Berlin 10587 Berlin, Germany

More information

Linear Model Selection and Regularization. especially usefull in high dimensions p>>100.

Linear Model Selection and Regularization. especially usefull in high dimensions p>>100. Linear Model Selection and Regularization especially usefull in high dimensions p>>100. 1 Why Linear Model Regularization? Linear models are simple, BUT consider p>>n, we have more features than data records

More information

ELEG Compressive Sensing and Sparse Signal Representations

ELEG Compressive Sensing and Sparse Signal Representations ELEG 867 - Compressive Sensing and Sparse Signal Representations Gonzalo R. Arce Depart. of Electrical and Computer Engineering University of Delaware Fall 211 Compressive Sensing G. Arce Fall, 211 1 /

More information

Signal Reconstruction from Sparse Representations: An Introdu. Sensing

Signal Reconstruction from Sparse Representations: An Introdu. Sensing Signal Reconstruction from Sparse Representations: An Introduction to Compressed Sensing December 18, 2009 Digital Data Acquisition Suppose we want to acquire some real world signal digitally. Applications

More information

r v i e w o f s o m e r e c e n t d e v e l o p m

r v i e w o f s o m e r e c e n t d e v e l o p m O A D O 4 7 8 O - O O A D OA 4 7 8 / D O O 3 A 4 7 8 / S P O 3 A A S P - * A S P - S - P - A S P - - - - L S UM 5 8 - - 4 3 8 -F 69 - V - F U 98F L 69V S U L S UM58 P L- SA L 43 ˆ UéL;S;UéL;SAL; - - -

More information

IMA Preprint Series # 2211

IMA Preprint Series # 2211 LEARNING TO SENSE SPARSE SIGNALS: SIMULTANEOUS SENSING MATRIX AND SPARSIFYING DICTIONARY OPTIMIZATION By Julio Martin Duarte-Carvajalino and Guillermo Sapiro IMA Preprint Series # 2211 ( May 2008 ) INSTITUTE

More information

Kristoffer H. Rose. Version /01/07

Kristoffer H. Rose. Version /01/07 Summary of qsymbols Kristoffer H. Rose Version 1.12 1997/01/07 Abstract qsymbols is a L A TEX [1] package defining systematic mnemonic abbreviations, starting with a single open quote for symbols, and

More information

Computer vision: models, learning and inference. Chapter 13 Image preprocessing and feature extraction

Computer vision: models, learning and inference. Chapter 13 Image preprocessing and feature extraction Computer vision: models, learning and inference Chapter 13 Image preprocessing and feature extraction Preprocessing The goal of pre-processing is to try to reduce unwanted variation in image due to lighting,

More information

Computational Physics PHYS 420

Computational Physics PHYS 420 Computational Physics PHYS 420 Dr Richard H. Cyburt Assistant Professor of Physics My office: 402c in the Science Building My phone: (304) 384-6006 My email: rcyburt@concord.edu My webpage: www.concord.edu/rcyburt

More information

Code Generation for Embedded Convex Optimization

Code Generation for Embedded Convex Optimization Code Generation for Embedded Convex Optimization Jacob Mattingley Stanford University October 2010 Convex optimization Problems solvable reliably and efficiently Widely used in scheduling, finance, engineering

More information

Sparse & Functional Principal Components Analysis

Sparse & Functional Principal Components Analysis Sparse & Functional Principal Components Analysis Genevera I. Allen Department of Statistics and Electrical and Computer Engineering, Rice University, Department of Pediatrics-Neurology, Baylor College

More information

Modern Methods of Data Analysis - WS 07/08

Modern Methods of Data Analysis - WS 07/08 Modern Methods of Data Analysis Lecture XV (04.02.08) Contents: Function Minimization (see E. Lohrmann & V. Blobel) Optimization Problem Set of n independent variables Sometimes in addition some constraints

More information

Case Study IV: Bayesian clustering of Alzheimer patients

Case Study IV: Bayesian clustering of Alzheimer patients Case Study IV: Bayesian clustering of Alzheimer patients Mike Wiper and Conchi Ausín Department of Statistics Universidad Carlos III de Madrid Advanced Statistics and Data Mining Summer School 2nd - 6th

More information

Introduction to Optimization

Introduction to Optimization Introduction to Optimization Second Order Optimization Methods Marc Toussaint U Stuttgart Planned Outline Gradient-based optimization (1st order methods) plain grad., steepest descent, conjugate grad.,

More information

Image reconstruction based on back propagation learning in Compressed Sensing theory

Image reconstruction based on back propagation learning in Compressed Sensing theory Image reconstruction based on back propagation learning in Compressed Sensing theory Gaoang Wang Project for ECE 539 Fall 2013 Abstract Over the past few years, a new framework known as compressive sampling

More information

Linear Methods for Regression and Shrinkage Methods

Linear Methods for Regression and Shrinkage Methods Linear Methods for Regression and Shrinkage Methods Reference: The Elements of Statistical Learning, by T. Hastie, R. Tibshirani, J. Friedman, Springer 1 Linear Regression Models Least Squares Input vectors

More information

Time Series Analysis by State Space Methods

Time Series Analysis by State Space Methods Time Series Analysis by State Space Methods Second Edition J. Durbin London School of Economics and Political Science and University College London S. J. Koopman Vrije Universiteit Amsterdam OXFORD UNIVERSITY

More information

Data Preprocessing. Javier Béjar. URL - Spring 2018 CS - MAI 1/78 BY: $\

Data Preprocessing. Javier Béjar. URL - Spring 2018 CS - MAI 1/78 BY: $\ Data Preprocessing Javier Béjar BY: $\ URL - Spring 2018 C CS - MAI 1/78 Introduction Data representation Unstructured datasets: Examples described by a flat set of attributes: attribute-value matrix Structured

More information

Adaptive Reconstruction Methods for Low-Dose Computed Tomography

Adaptive Reconstruction Methods for Low-Dose Computed Tomography Adaptive Reconstruction Methods for Low-Dose Computed Tomography Joseph Shtok Ph.D. supervisors: Prof. Michael Elad, Dr. Michael Zibulevsky. Technion IIT, Israel, 011 Ph.D. Talk, Apr. 01 Contents of this

More information

INLA: Integrated Nested Laplace Approximations

INLA: Integrated Nested Laplace Approximations INLA: Integrated Nested Laplace Approximations John Paige Statistics Department University of Washington October 10, 2017 1 The problem Markov Chain Monte Carlo (MCMC) takes too long in many settings.

More information

Topic 4: Vectors Markscheme 4.6 Intersection of Lines and Planes Paper 2

Topic 4: Vectors Markscheme 4.6 Intersection of Lines and Planes Paper 2 Topic : Vectors Markscheme. Intersection of Lines and Planes Paper. Using an elimination method, x y + z x y z x y x + y 8 y Therefore x, y, z Using matrices, x y z x y z 5 (using a graphic display calculator)

More information

CoxFlexBoost: Fitting Structured Survival Models

CoxFlexBoost: Fitting Structured Survival Models CoxFlexBoost: Fitting Structured Survival Models Benjamin Hofner 1 Institut für Medizininformatik, Biometrie und Epidemiologie (IMBE) Friedrich-Alexander-Universität Erlangen-Nürnberg joint work with Torsten

More information

GAMs semi-parametric GLMs. Simon Wood Mathematical Sciences, University of Bath, U.K.

GAMs semi-parametric GLMs. Simon Wood Mathematical Sciences, University of Bath, U.K. GAMs semi-parametric GLMs Simon Wood Mathematical Sciences, University of Bath, U.K. Generalized linear models, GLM 1. A GLM models a univariate response, y i as g{e(y i )} = X i β where y i Exponential

More information

Model selection and validation 1: Cross-validation

Model selection and validation 1: Cross-validation Model selection and validation 1: Cross-validation Ryan Tibshirani Data Mining: 36-462/36-662 March 26 2013 Optional reading: ISL 2.2, 5.1, ESL 7.4, 7.10 1 Reminder: modern regression techniques Over the

More information

Lecture 24: Generalized Additive Models Stat 704: Data Analysis I, Fall 2010

Lecture 24: Generalized Additive Models Stat 704: Data Analysis I, Fall 2010 Lecture 24: Generalized Additive Models Stat 704: Data Analysis I, Fall 2010 Tim Hanson, Ph.D. University of South Carolina T. Hanson (USC) Stat 704: Data Analysis I, Fall 2010 1 / 26 Additive predictors

More information

Retrieval by Content. Part 3: Text Retrieval Latent Semantic Indexing. Srihari: CSE 626 1

Retrieval by Content. Part 3: Text Retrieval Latent Semantic Indexing. Srihari: CSE 626 1 Retrieval by Content art 3: Text Retrieval Latent Semantic Indexing Srihari: CSE 626 1 Latent Semantic Indexing LSI isadvantage of exclusive use of representing a document as a T-dimensional vector of

More information

2. Unlock the Customization Features: The Edit Button Click the "Edit" button on the Dashboard Home Page to unlock the customization features.

2. Unlock the Customization Features: The Edit Button Click the Edit button on the Dashboard Home Page to unlock the customization features. 1. Introduction: Customize Content for Your Students The Garland Science Learning System (GSLS) is completely customizable. You can edit, delete, or add content to create a unique learning experience for

More information

Learning Algorithms for Medical Image Analysis. Matteo Santoro slipguru

Learning Algorithms for Medical Image Analysis. Matteo Santoro slipguru Learning Algorithms for Medical Image Analysis Matteo Santoro slipguru santoro@disi.unige.it June 8, 2010 Outline 1. learning-based strategies for quantitative image analysis 2. automatic annotation of

More information

Overfitting. Machine Learning CSE546 Carlos Guestrin University of Washington. October 2, Bias-Variance Tradeoff

Overfitting. Machine Learning CSE546 Carlos Guestrin University of Washington. October 2, Bias-Variance Tradeoff Overfitting Machine Learning CSE546 Carlos Guestrin University of Washington October 2, 2013 1 Bias-Variance Tradeoff Choice of hypothesis class introduces learning bias More complex class less bias More

More information

Robust Kernel Methods in Clustering and Dimensionality Reduction Problems

Robust Kernel Methods in Clustering and Dimensionality Reduction Problems Robust Kernel Methods in Clustering and Dimensionality Reduction Problems Jian Guo, Debadyuti Roy, Jing Wang University of Michigan, Department of Statistics Introduction In this report we propose robust

More information

FMA901F: Machine Learning Lecture 3: Linear Models for Regression. Cristian Sminchisescu

FMA901F: Machine Learning Lecture 3: Linear Models for Regression. Cristian Sminchisescu FMA901F: Machine Learning Lecture 3: Linear Models for Regression Cristian Sminchisescu Machine Learning: Frequentist vs. Bayesian In the frequentist setting, we seek a fixed parameter (vector), with value(s)

More information

A brief description of the particle finite element method (PFEM2). Extensions to free surface

A brief description of the particle finite element method (PFEM2). Extensions to free surface A brief description of the particle finite element method (PFEM2). Extensions to free surface flows. Juan M. Gimenez, L.M. González, CIMEC Universidad Nacional del Litoral (UNL) Santa Fe, Argentina Universidad

More information

Problem 1 (20 pt) Answer the following questions, and provide an explanation for each question.

Problem 1 (20 pt) Answer the following questions, and provide an explanation for each question. Problem 1 Answer the following questions, and provide an explanation for each question. (5 pt) Can linear regression work when all X values are the same? When all Y values are the same? (5 pt) Can linear

More information

Data Preprocessing. Javier Béjar AMLT /2017 CS - MAI. (CS - MAI) Data Preprocessing AMLT / / 71 BY: $\

Data Preprocessing. Javier Béjar AMLT /2017 CS - MAI. (CS - MAI) Data Preprocessing AMLT / / 71 BY: $\ Data Preprocessing S - MAI AMLT - 2016/2017 (S - MAI) Data Preprocessing AMLT - 2016/2017 1 / 71 Outline 1 Introduction Data Representation 2 Data Preprocessing Outliers Missing Values Normalization Discretization

More information

Wavelet Neural Networks

Wavelet Neural Networks Wavelet Neural Networks and their application in the study of dynamical systems David Veitch Dissertation submitted for the MSc in Data Analysis, Networks and Nonlinear Dynamics. Department of Mathematics

More information

AN ADDITIVE BIVARIATE HIERARCHICAL MODEL FOR FUNCTIONAL DATA AND RELATED COMPUTATIONS. A Dissertation ANDREW MIDDLETON REDD

AN ADDITIVE BIVARIATE HIERARCHICAL MODEL FOR FUNCTIONAL DATA AND RELATED COMPUTATIONS. A Dissertation ANDREW MIDDLETON REDD AN ADDITIVE BIVARIATE HIERARCHICAL MODEL FOR FUNCTIONAL DATA AND RELATED COMPUTATIONS A Dissertation by ANDREW MIDDLETON REDD Submitted to the Office of Graduate Studies of Texas A&M University in partial

More information

STAT 705 Introduction to generalized additive models

STAT 705 Introduction to generalized additive models STAT 705 Introduction to generalized additive models Timothy Hanson Department of Statistics, University of South Carolina Stat 705: Data Analysis II 1 / 22 Generalized additive models Consider a linear

More information

Assessing the Quality of the Natural Cubic Spline Approximation

Assessing the Quality of the Natural Cubic Spline Approximation Assessing the Quality of the Natural Cubic Spline Approximation AHMET SEZER ANADOLU UNIVERSITY Department of Statisticss Yunus Emre Kampusu Eskisehir TURKEY ahsst12@yahoo.com Abstract: In large samples,

More information

CSE 250B Project Assignment 4

CSE 250B Project Assignment 4 CSE 250B Project Assignment 4 Hani Altwary haltwa@cs.ucsd.edu Kuen-Han Lin kul016@ucsd.edu Toshiro Yamada toyamada@ucsd.edu Abstract The goal of this project is to implement the Semi-Supervised Recursive

More information

INTRODUCTION TO CLUSTER ALGEBRAS

INTRODUCTION TO CLUSTER ALGEBRAS INTRODUCTION TO CLUSTER ALGEBRAS NAN LI (MIT) Cluster algebras are a class of commutative ring, introduced in 000 by Fomin and Zelevinsky, originally to study Lusztig s dual canonical basis and total positivity.

More information

Parameterization of triangular meshes

Parameterization of triangular meshes Parameterization of triangular meshes Michael S. Floater November 10, 2009 Triangular meshes are often used to represent surfaces, at least initially, one reason being that meshes are relatively easy to

More information

MODERN FACTOR ANALYSIS

MODERN FACTOR ANALYSIS MODERN FACTOR ANALYSIS Harry H. Harman «ö THE pigj UNIVERSITY OF CHICAGO PRESS Contents LIST OF ILLUSTRATIONS GUIDE TO NOTATION xv xvi Parti Foundations of Factor Analysis 1. INTRODUCTION 3 1.1. Brief

More information

Bayesian Model Averaging over Directed Acyclic Graphs With Implications for Prediction in Structural Equation Modeling

Bayesian Model Averaging over Directed Acyclic Graphs With Implications for Prediction in Structural Equation Modeling ing over Directed Acyclic Graphs With Implications for Prediction in ing David Kaplan Department of Educational Psychology Case April 13th, 2015 University of Nebraska-Lincoln 1 / 41 ing Case This work

More information

Splines and penalized regression

Splines and penalized regression Splines and penalized regression November 23 Introduction We are discussing ways to estimate the regression function f, where E(y x) = f(x) One approach is of course to assume that f has a certain shape,

More information

Sparsity Based Regularization

Sparsity Based Regularization 9.520: Statistical Learning Theory and Applications March 8th, 200 Sparsity Based Regularization Lecturer: Lorenzo Rosasco Scribe: Ioannis Gkioulekas Introduction In previous lectures, we saw how regularization

More information

A popular method for moving beyond linearity. 2. Basis expansion and regularization 1. Examples of transformations. Piecewise-polynomials and splines

A popular method for moving beyond linearity. 2. Basis expansion and regularization 1. Examples of transformations. Piecewise-polynomials and splines A popular method for moving beyond linearity 2. Basis expansion and regularization 1 Idea: Augment the vector inputs x with additional variables which are transformation of x use linear models in this

More information

TECHNICAL REPORT NO December 11, 2001

TECHNICAL REPORT NO December 11, 2001 DEPARTMENT OF STATISTICS University of Wisconsin 2 West Dayton St. Madison, WI 5376 TECHNICAL REPORT NO. 48 December, 2 Penalized Log Likelihood Density Estimation, via Smoothing-Spline ANOVA and rangacv

More information

Physics 736. Experimental Methods in Nuclear-, Particle-, and Astrophysics. - Statistical Methods -

Physics 736. Experimental Methods in Nuclear-, Particle-, and Astrophysics. - Statistical Methods - Physics 736 Experimental Methods in Nuclear-, Particle-, and Astrophysics - Statistical Methods - Karsten Heeger heeger@wisc.edu Course Schedule and Reading course website http://neutrino.physics.wisc.edu/teaching/phys736/

More information

Theoretical Concepts of Machine Learning

Theoretical Concepts of Machine Learning Theoretical Concepts of Machine Learning Part 2 Institute of Bioinformatics Johannes Kepler University, Linz, Austria Outline 1 Introduction 2 Generalization Error 3 Maximum Likelihood 4 Noise Models 5

More information

Unsupervised Learning

Unsupervised Learning Unsupervised Learning Learning without Class Labels (or correct outputs) Density Estimation Learn P(X) given training data for X Clustering Partition data into clusters Dimensionality Reduction Discover

More information

Exact Models for Open Field Layout Problem with l 2 and l 1 Distances

Exact Models for Open Field Layout Problem with l 2 and l 1 Distances R u t c o r Research R e p o r t Exact Models for Open Field Layout Problem with l 2 and l 1 Distances Gergely Kovács a Béla Vizvári b RRR 1-2014, May 2014, RUTCOR Rutgers Center for Operations Research

More information

Extended target tracking using PHD filters

Extended target tracking using PHD filters Ulm University 2014 01 29 1(35) With applications to video data and laser range data Division of Automatic Control Department of Electrical Engineering Linöping university Linöping, Sweden Presentation

More information

ENHANCED MONITORING USING MULTISCALE EXPONENTIALLY WEIGHTED MOVING AVERAGE CONTROL CHARTS

ENHANCED MONITORING USING MULTISCALE EXPONENTIALLY WEIGHTED MOVING AVERAGE CONTROL CHARTS ENHANCED MONITORING USING MULTISCALE EXPONENTIALLY WEIGHTED MOVING AVERAGE CONTROL CHARTS A Thesis by MD. ALAMGIR MOJIBUL HAQUE Submitted to the Office of Graduate and Professional Studies of Texas A&M

More information

Short on camera geometry and camera calibration

Short on camera geometry and camera calibration Short on camera geometry and camera calibration Maria Magnusson, maria.magnusson@liu.se Computer Vision Laboratory, Department of Electrical Engineering, Linköping University, Sweden Report No: LiTH-ISY-R-3070

More information

A Neuro Probabilistic Language Model Bengio et. al. 2003

A Neuro Probabilistic Language Model Bengio et. al. 2003 A Neuro Probabilistic Language Model Bengio et. al. 2003 Class Discussion Notes Scribe: Olivia Winn February 1, 2016 Opening thoughts (or why this paper is interesting): Word embeddings currently have

More information

Harmonic Spline Series Representation of Scaling Functions

Harmonic Spline Series Representation of Scaling Functions Harmonic Spline Series Representation of Scaling Functions Thierry Blu and Michael Unser Biomedical Imaging Group, STI/BIO-E, BM 4.34 Swiss Federal Institute of Technology, Lausanne CH-5 Lausanne-EPFL,

More information

CS321 Introduction To Numerical Methods

CS321 Introduction To Numerical Methods CS3 Introduction To Numerical Methods Fuhua (Frank) Cheng Department of Computer Science University of Kentucky Lexington KY 456-46 - - Table of Contents Errors and Number Representations 3 Error Types

More information

Contents. I Basics 1. Copyright by SIAM. Unauthorized reproduction of this article is prohibited.

Contents. I Basics 1. Copyright by SIAM. Unauthorized reproduction of this article is prohibited. page v Preface xiii I Basics 1 1 Optimization Models 3 1.1 Introduction... 3 1.2 Optimization: An Informal Introduction... 4 1.3 Linear Equations... 7 1.4 Linear Optimization... 10 Exercises... 12 1.5

More information

Chapter 18. Geometric Operations

Chapter 18. Geometric Operations Chapter 18 Geometric Operations To this point, the image processing operations have computed the gray value (digital count) of the output image pixel based on the gray values of one or more input pixels;

More information

Modeling Criminal Careers as Departures From a Unimodal Population Age-Crime Curve: The Case of Marijuana Use

Modeling Criminal Careers as Departures From a Unimodal Population Age-Crime Curve: The Case of Marijuana Use Modeling Criminal Careers as Departures From a Unimodal Population Curve: The Case of Marijuana Use Donatello Telesca, Elena A. Erosheva, Derek A. Kreader, & Ross Matsueda April 15, 2014 extends Telesca

More information

Efficient Iterative LP Decoding of LDPC Codes with Alternating Direction Method of Multipliers

Efficient Iterative LP Decoding of LDPC Codes with Alternating Direction Method of Multipliers Efficient Iterative LP Decoding of LDPC Codes with Alternating Direction Method of Multipliers Xiaojie Zhang Samsung R&D America, Dallas, Texas 758 Email: eric.zhang@samsung.com Paul H. Siegel University

More information

Preface to the Second Edition. Preface to the First Edition. 1 Introduction 1

Preface to the Second Edition. Preface to the First Edition. 1 Introduction 1 Preface to the Second Edition Preface to the First Edition vii xi 1 Introduction 1 2 Overview of Supervised Learning 9 2.1 Introduction... 9 2.2 Variable Types and Terminology... 9 2.3 Two Simple Approaches

More information

Machine Learning (BSMC-GA 4439) Wenke Liu

Machine Learning (BSMC-GA 4439) Wenke Liu Machine Learning (BSMC-GA 4439) Wenke Liu 01-31-017 Outline Background Defining proximity Clustering methods Determining number of clusters Comparing two solutions Cluster analysis as unsupervised Learning

More information

Diffusion Wavelets for Natural Image Analysis

Diffusion Wavelets for Natural Image Analysis Diffusion Wavelets for Natural Image Analysis Tyrus Berry December 16, 2011 Contents 1 Project Description 2 2 Introduction to Diffusion Wavelets 2 2.1 Diffusion Multiresolution............................

More information

CHAPTER 1 INTRODUCTION

CHAPTER 1 INTRODUCTION Introduction CHAPTER 1 INTRODUCTION Mplus is a statistical modeling program that provides researchers with a flexible tool to analyze their data. Mplus offers researchers a wide choice of models, estimators,

More information

3 Interior Point Method

3 Interior Point Method 3 Interior Point Method Linear programming (LP) is one of the most useful mathematical techniques. Recent advances in computer technology and algorithms have improved computational speed by several orders

More information

Apprenticeship Learning for Reinforcement Learning. with application to RC helicopter flight Ritwik Anand, Nick Haliday, Audrey Huang

Apprenticeship Learning for Reinforcement Learning. with application to RC helicopter flight Ritwik Anand, Nick Haliday, Audrey Huang Apprenticeship Learning for Reinforcement Learning with application to RC helicopter flight Ritwik Anand, Nick Haliday, Audrey Huang Table of Contents Introduction Theory Autonomous helicopter control

More information

Monte Carlo for Spatial Models

Monte Carlo for Spatial Models Monte Carlo for Spatial Models Murali Haran Department of Statistics Penn State University Penn State Computational Science Lectures April 2007 Spatial Models Lots of scientific questions involve analyzing

More information

WESTFÄLISCHE WILHELMS-UNIVERSITÄT MÜNSTER

WESTFÄLISCHE WILHELMS-UNIVERSITÄT MÜNSTER MÜNSTER Adaptive Discretization of Liftings for Curvature Regularization SIAM Conference on Imaging Science 2016, Albuquerque, NM Ulrich Hartleif, Prof. Dr. Benedikt Wirth May 23, 2016 Outline MÜNSTER

More information

Detection Performance of Radar Compressive Sensing in Noisy Environments

Detection Performance of Radar Compressive Sensing in Noisy Environments Detection Performance of Radar Compressive Sensing in Noisy Environments Asmita Korde a,damon Bradley b and Tinoosh Mohsenin a a Department of Computer Science and Electrical Engineering, University of

More information

Data Analysis 3. Support Vector Machines. Jan Platoš October 30, 2017

Data Analysis 3. Support Vector Machines. Jan Platoš October 30, 2017 Data Analysis 3 Support Vector Machines Jan Platoš October 30, 2017 Department of Computer Science Faculty of Electrical Engineering and Computer Science VŠB - Technical University of Ostrava Table of

More information

Machine Learning Techniques for Detecting Hierarchical Interactions in GLM s for Insurance Premiums

Machine Learning Techniques for Detecting Hierarchical Interactions in GLM s for Insurance Premiums Machine Learning Techniques for Detecting Hierarchical Interactions in GLM s for Insurance Premiums José Garrido Department of Mathematics and Statistics Concordia University, Montreal EAJ 2016 Lyon, September

More information

Clustering K-means. Machine Learning CSEP546 Carlos Guestrin University of Washington February 18, Carlos Guestrin

Clustering K-means. Machine Learning CSEP546 Carlos Guestrin University of Washington February 18, Carlos Guestrin Clustering K-means Machine Learning CSEP546 Carlos Guestrin University of Washington February 18, 2014 Carlos Guestrin 2005-2014 1 Clustering images Set of Images [Goldberger et al.] Carlos Guestrin 2005-2014

More information

Edge and local feature detection - 2. Importance of edge detection in computer vision

Edge and local feature detection - 2. Importance of edge detection in computer vision Edge and local feature detection Gradient based edge detection Edge detection by function fitting Second derivative edge detectors Edge linking and the construction of the chain graph Edge and local feature

More information

Numerical Methods for PDEs : Video 11: 1D FiniteFebruary Difference 15, Mappings Theory / 15

Numerical Methods for PDEs : Video 11: 1D FiniteFebruary Difference 15, Mappings Theory / 15 22.520 Numerical Methods for PDEs : Video 11: 1D Finite Difference Mappings Theory and Matlab February 15, 2015 22.520 Numerical Methods for PDEs : Video 11: 1D FiniteFebruary Difference 15, Mappings 2015

More information

62 Basilio Bona - Dynamic Modelling

62 Basilio Bona - Dynamic Modelling 6 Basilio Bona - Dynamic Modelling Only 6 multiplications are required compared to the 7 that are involved in the product between two 3 3 rotation matrices...6 Quaternions Quaternions were introduced by

More information

CSE 547: Machine Learning for Big Data Spring Problem Set 2. Please read the homework submission policies.

CSE 547: Machine Learning for Big Data Spring Problem Set 2. Please read the homework submission policies. CSE 547: Machine Learning for Big Data Spring 2019 Problem Set 2 Please read the homework submission policies. 1 Principal Component Analysis and Reconstruction (25 points) Let s do PCA and reconstruct

More information

Package mcemglm. November 29, 2015

Package mcemglm. November 29, 2015 Type Package Package mcemglm November 29, 2015 Title Maximum Likelihood Estimation for Generalized Linear Mixed Models Version 1.1 Date 2015-11-28 Author Felipe Acosta Archila Maintainer Maximum likelihood

More information

Chap.12 Kernel methods [Book, Chap.7]

Chap.12 Kernel methods [Book, Chap.7] Chap.12 Kernel methods [Book, Chap.7] Neural network methods became popular in the mid to late 1980s, but by the mid to late 1990s, kernel methods have also become popular in machine learning. The first

More information

COMP 558 lecture 19 Nov. 17, 2010

COMP 558 lecture 19 Nov. 17, 2010 COMP 558 lecture 9 Nov. 7, 2 Camera calibration To estimate the geometry of 3D scenes, it helps to know the camera parameters, both external and internal. The problem of finding all these parameters is

More information

arxiv: v1 [stat.ap] 1 May 2018

arxiv: v1 [stat.ap] 1 May 2018 A Discrete View of the Indian Monsoon to Identify Spatial Patterns of Rainfall Adway Mitra 1, Amit Apte 2, Rama Govindarajan 2, Vishal Vasan 2, and Sreekar Vadlamani 3,4 arxiv:1805.00414v1 [stat.ap] 1

More information

Using DIC to compare selection models with non-ignorable missing responses

Using DIC to compare selection models with non-ignorable missing responses Using DIC to compare selection models with non-ignorable missing responses Abstract Data with missing responses generated by a non-ignorable missingness mechanism can be analysed by jointly modelling the

More information

A MRF Shape Prior for Facade Parsing with Occlusions Supplementary Material

A MRF Shape Prior for Facade Parsing with Occlusions Supplementary Material A MRF Shape Prior for Facade Parsing with Occlusions Supplementary Material Mateusz Koziński, Raghudeep Gadde, Sergey Zagoruyko, Guillaume Obozinski and Renaud Marlet Université Paris-Est, LIGM (UMR CNRS

More information

Diffuse Optical Tomography, Inverse Problems, and Optimization. Mary Katherine Huffman. Undergraduate Research Fall 2011 Spring 2012

Diffuse Optical Tomography, Inverse Problems, and Optimization. Mary Katherine Huffman. Undergraduate Research Fall 2011 Spring 2012 Diffuse Optical Tomography, Inverse Problems, and Optimization Mary Katherine Huffman Undergraduate Research Fall 11 Spring 12 1. Introduction. This paper discusses research conducted in order to investigate

More information

Mixed Effects Models. Biljana Jonoska Stojkova Applied Statistics and Data Science Group (ASDa) Department of Statistics, UBC.

Mixed Effects Models. Biljana Jonoska Stojkova Applied Statistics and Data Science Group (ASDa) Department of Statistics, UBC. Mixed Effects Models Biljana Jonoska Stojkova Applied Statistics and Data Science Group (ASDa) Department of Statistics, UBC March 6, 2018 Resources for statistical assistance Department of Statistics

More information

Bayesian Spherical Wavelet Shrinkage: Applications to Shape Analysis

Bayesian Spherical Wavelet Shrinkage: Applications to Shape Analysis Bayesian Spherical Wavelet Shrinkage: Applications to Shape Analysis Xavier Le Faucheur a, Brani Vidakovic b and Allen Tannenbaum a a School of Electrical and Computer Engineering, b Department of Biomedical

More information

Dealing with Categorical Data Types in a Designed Experiment

Dealing with Categorical Data Types in a Designed Experiment Dealing with Categorical Data Types in a Designed Experiment Part II: Sizing a Designed Experiment When Using a Binary Response Best Practice Authored by: Francisco Ortiz, PhD STAT T&E COE The goal of

More information

The Pre-Image Problem in Kernel Methods

The Pre-Image Problem in Kernel Methods The Pre-Image Problem in Kernel Methods James Kwok Ivor Tsang Department of Computer Science Hong Kong University of Science and Technology Hong Kong The Pre-Image Problem in Kernel Methods ICML-2003 1

More information