Journal Club Journal of Chemometrics May 2010 August 23, 2010.

Slides:



Advertisements
Similar presentations
Zhen Lu CPACT University of Newcastle MDC Technology Reduced Hessian Sequential Quadratic Programming(SQP)
Advertisements

Eigen Decomposition and Singular Value Decomposition
Self-Organizing Maps Projection of p dimensional observations to a two (or one) dimensional grid space Constraint version of K-means clustering –Prototypes.
Principal Component Analysis Based on L1-Norm Maximization Nojun Kwak IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008.
Dimension reduction (2) Projection pursuit ICA NCA Partial Least Squares Blais. “The role of the environment in synaptic plasticity…..” (1998) Liao et.
Use of Kalman filters in time and frequency analysis John Davis 1st May 2011.
1er. Escuela Red ProTIC - Tandil, de Abril, 2006 Principal component analysis (PCA) is a technique that is useful for the compression and classification.
Aspects of Conditional Simulation and estimation of hydraulic conductivity in coastal aquifers" Luit Jan Slooten.
Response Surface Method Principle Component Analysis
Bayesian Robust Principal Component Analysis Presenter: Raghu Ranganathan ECE / CMR Tennessee Technological University January 21, 2011 Reading Group (Xinghao.
1cs542g-term High Dimensional Data  So far we’ve considered scalar data values f i (or interpolated/approximated each component of vector values.
Principal Component Analysis CMPUT 466/551 Nilanjan Ray.
Dimensionality reduction. Outline From distances to points : – MultiDimensional Scaling (MDS) – FastMap Dimensionality Reductions or data projections.
Principal Component Analysis
3D Geometry for Computer Graphics
Factor Analysis Purpose of Factor Analysis
Some useful linear algebra. Linearly independent vectors span(V): span of vector space V is all linear combinations of vectors v i, i.e.
L15:Microarray analysis (Classification) The Biological Problem Two conditions that need to be differentiated, (Have different treatments). EX: ALL (Acute.
Principal component analysis (PCA)
An Introduction to Kernel-Based Learning Algorithms K.-R. Muller, S. Mika, G. Ratsch, K. Tsuda and B. Scholkopf Presented by: Joanna Giforos CS8980: Topics.
Dimensional reduction, PCA
Prénom Nom Document Analysis: Data Analysis and Clustering Prof. Rolf Ingold, University of Fribourg Master course, spring semester 2008.
CHAPTER 19 Correspondence Analysis From: McCune, B. & J. B. Grace Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, Oregon.
Independent Component Analysis (ICA) and Factor Analysis (FA)
The Terms that You Have to Know! Basis, Linear independent, Orthogonal Column space, Row space, Rank Linear combination Linear transformation Inner product.
3D Geometry for Computer Graphics
Principal Components Analysis (PCA) 273A Intro Machine Learning.
Ordinary least squares regression (OLS)
Continuous Latent Variables --Bishop
A Unifying Review of Linear Gaussian Models
Techniques for studying correlation and covariance structure
Survey on ICA Technical Report, Aapo Hyvärinen, 1999.
Robust PCA in Stata Vincenzo Verardi FUNDP (Namur) and ULB (Brussels), Belgium FNRS Associate Researcher.
Summarized by Soo-Jin Kim
Principle Component Analysis (PCA) Networks (§ 5.8) PCA: a statistical procedure –Reduce dimensionality of input vectors Too many features, some of them.
Linear Least Squares Approximation. 2 Definition (point set case) Given a point set x 1, x 2, …, x n  R d, linear least squares fitting amounts to find.
Chapter 2 Dimensionality Reduction. Linear Methods
Presented By Wanchen Lu 2/25/2013
Principal Components Analysis BMTRY 726 3/27/14. Uses Goal: Explain the variability of a set of variables using a “small” set of linear combinations of.
1 Hybrid methods for solving large-scale parameter estimation problems Carlos A. Quintero 1 Miguel Argáez 1 Hector Klie 2 Leticia Velázquez 1 Mary Wheeler.
Chapter 15 Modeling of Data. Statistics of Data Mean (or average): Variance: Median: a value x j such that half of the data are bigger than it, and half.
Additive Data Perturbation: data reconstruction attacks.
EMIS 8381 – Spring Netflix and Your Next Movie Night Nonlinear Programming Ron Andrews EMIS 8381.
Principal Component Analysis Bamshad Mobasher DePaul University Bamshad Mobasher DePaul University.
Orthogonality and Least Squares
© 2011 Autodesk Freely licensed for use by educational institutions. Reuse and changes require a note indicating that content has been modified from the.
N– variate Gaussian. Some important characteristics: 1)The pdf of n jointly Gaussian R.V.’s is completely described by means, variances and covariances.
Colorado Center for Astrodynamics Research The University of Colorado 1 STATISTICAL ORBIT DETERMINATION ASEN 5070 LECTURE 11 9/16,18/09.
A Flexible New Technique for Camera Calibration Zhengyou Zhang Sung Huh CSPS 643 Individual Presentation 1 February 25,
Tetris Agent Optimization Using Harmony Search Algorithm
Tony Jebara, Columbia University Advanced Machine Learning & Perception Instructor: Tony Jebara.
Reduces time complexity: Less computation Reduces space complexity: Less parameters Simpler models are more robust on small datasets More interpretable;
EIGENSYSTEMS, SVD, PCA Big Data Seminar, Dedi Gadot, December 14 th, 2014.
Principal Component Analysis (PCA)
ECE 530 – Analysis Techniques for Large-Scale Electrical Systems Prof. Hao Zhu Dept. of Electrical and Computer Engineering University of Illinois at Urbana-Champaign.
Principal Component Analysis Zelin Jia Shengbin Lin 10/20/2015.
Feature Extraction 主講人:虞台文. Content Principal Component Analysis (PCA) PCA Calculation — for Fewer-Sample Case Factor Analysis Fisher’s Linear Discriminant.
Part 3: Estimation of Parameters. Estimation of Parameters Most of the time, we have random samples but not the densities given. If the parametric form.
Background on Classification
Principle Component Analysis (PCA) Networks (§ 5.8)
Some useful linear algebra
Probabilistic Models with Latent Variables
Principal Component Analysis
Orthogonality and Least Squares
Presented by Nagesh Adluru
Feature space tansformation methods
Generally Discriminant Analysis
Orthogonality and Least Squares
Marios Mattheakis and Pavlos Protopapas
Outline Variance Matrix of Stochastic Variables and Orthogonal Transforms Principle Component Analysis Generalized Eigenvalue Decomposition.
Presentation transcript:

Journal Club Journal of Chemometrics May 2010 August 23, 2010

An efficient nonlinear programming strategy for PCA models with incomplete data sets Rodrigo López-Negrete de la Fuentea, Salvador García-Muñozb and Lorenz T. Biegler J. Chemometrics 2010; 24: 301–311

Questions addressed: – How to obtain the parameters of PCA models in the presence of incomplete data sets based in non-linear programming strategy. – How nonlinear programming approach is better suited when there are large amounts of missing values.

Methods PCA with full data-set: – Given: where, T: Scores(projection) P:Loading Rx:Residulas Problem 1: Solution: the solution will be given by the largest eigenvalue of the covariance matrix of X. Largest eigen values: largest variance variance of XX’ Problem 2: Solution: Has the same form as that of the solution of the maximization Problem.

Methods PCA with full data-set: Using SVD: Two problems have the same solution.

Methods PCA with full data-set: Principal Components via NIPALS algorithm:

Methods PCA with incomplete data-set: Taking gradient, wrt t and p

Methods PCA with incomplete data-set: Principal Components via modified NIPALS algorithm:

Methods PCA with incomplete data-set: - X is the matrix of data where the missing elements have been zeroed out. - Constraint (20b) forces the loadings to be orthonormal. - Constraint (20c) makes the score vectors orthogonal to each other. - Constraint (20d) forces the scores to have zero mean. - It is clear that if there are no missing values problem (20) will reduce to problem (4) (min problem) for the first a principal components. Let Yi,j = Xi,j + Zi,j where Xi,j are the values of the data that are equal to zero for the missing elements, and Zi,j are the imputed values that should be zero for the nonmissing elements.

Methods PCA with incomplete data-set: constrained problem will be solved directly, the scores and loadings obtained with the NLP will be orthogonal as needed by the PCA model, which is not true for the modified NIPALS.

RESULTS Numerical simulations were done by generating a data set with 1000 rows and 100 columns from a known four-dimensional latent space with added random Gaussian noise. Values were then removed to generate data sets with missing value percentages ranging from 1 to 70%.

RESULTS

Industrial Example: - Data from 76 common pharmaceutical materials were made available (Pfizer Inc.) and the data span over 10 years of testing. Due to the reasons outlined above, approximately 61% of the data were missing. - For this example, three principal components were used in all models due to: a sudden drop in the eigenvalue for the fourth component (from 13 to 2) for the NIPALS model and the very low percent of the total variance for the fourth component (1.4%) in the NLP.

Conclusion The NLP solutions take less time and iterations than the current state-of the-art algorithms, while still satisfying the constraints of the PCA model. The current platform allows the potential inclusion of a large number of observations that otherwise would be excluded from the model building exercise, still yielding a robust model with desirable properties. In the presence of large amounts of missing data, this method reduces the computational time (and number of iterations) required to calculate them.