Search

The gold standard for modeling multiple indicator measurement data is confirmatory factor analysis (CFA), which has many statistical advantages over traditional exploratory factor analysis (EFA). In most CFA applications, items are assumed to be pure indicators of the construct they intend to measure. However, despite our best efforts, this is often not the case. Cross-loadings incorrectly set to zero can only be expressed through the correlations between the factors, leading to biased factor correlations and to biased structural (regression) parameter estimates. This article introduces a third approach, which has emerged in the psychometric literature, viz., unrestricted factor analysis (UFA). UFA borrows strengths from both traditional EFA and CFA. In simulation studies, we show that ignoring cross-loadings even as low as .2 can substantially bias factor correlations when CFA is used and that even the commonly used guideline RMSEA [less than or equal to] .05 may be too lenient to guard against non-negligible bias in factor correlations in CFA. Next, we present two empirical applications using Schwartz's value theory, and electronic service quality. In the first case, UFA leads to much better model fit and more plausible regression estimates. In the second case, the difference is less dramatic but nevertheless, UFA provides richer results. We provide recommendations on when to use UFA vs. CFA., Author(s): Jan-Benedict E. M. Steenkamp [sup.1] , Alberto Maydeu-Olivares [sup.2] Author Affiliations: (1) grid.10698.36, 0000000122483208, University of North Carolina at Chapel Hill, , Campus Box 3490, McColl Building, 27599-3490, Chapel [...]

Most research on feigning orgasm has focused exclusively on women and on potential predictors of this behaviour, with little attention given to the underlying motives for doing so. There are [...]

This paper provides a methodological analysis of credit risk in momflacturing firms by using two different credit scoring approaches. The .first is the traditional discriminant approach jr bankruptcy prediction based [...]

Interface management (IM) is one of the major keys for construction project success. The severity of interface problems for different projects does not only delay the project, but also impacts overall project performance. This paper is an extension of a previous work that defined major IM problems in Alberta's construction projects. This research study intended to investigate, identify, and classify interface problem factors in Alberta's construction projects. The study included four stages. The first stage was a comprehensive literature review, pilot studies and face-to-face interviews in industry. In the second phase, a web-page questionnaire was conducted with participants from industry. Based on that, in the last two phases, a factor analysis and Pearson's correlation matrix were applied on the collected data. The study identified six IM factors, namely: 'management', 'information, bidding and contracting', 'by-law and regulation', 'technical engineering and site issues', and 'other interface problems'. Finally, correlation between IM factors and different construction data was tested. The data analysis results provided a comprehensive view of the main causes behind IM conflicts in Alberta's construction industry. Key words: interface management, construction, performance, factor analysis, correlation. La gestion des interfaces est l'une des cles de la reussite d'un projet de construction. La gravite des problemes d'interface pour les divers projets non seulement retarde le projet mais aussi affecte le rendement du projet dans son ensemble. Le present article est la suite d'un travail qui a defini les principaux problemes de gestion des interfaces dans les projets de construction en Alberta. La presente recherche visait a examiner, a identifier et a classifier les facteurs de problemes d'interface dans les projets de construction en Alberta. L'etude comprend quatre etapes. La premiere etape etait une revue complete de la litterature, des etudes pilotes et des entrevues en personne dans l'industrie. Dans la seconde etape, un questionnaire Web a ete rempli par des participants de l'industrie. Les deux dernieres etapes consistaient en une analyse factorielle et une matrice de correlation de Pearson, appliquees aux donnees colligees. L'etude identifiait six facteurs de gestion des interfaces, dont la <>, l'information, l'appel d'offre et la passation de contrat',les reglements administratifs et la reglementation>>, la technogenie et les questions d'emplacement'etles autres problemes d'interface>>. Finalement, la correlation entre les facteurs de gestion des interfaces et les diverses donnees de construction a ete verifiee. Les resultats de l'analyse des donnees ont fourni une vue de l'ensemble des principales causes dans les conflits de gestion des interfaces dans l'industrie de la construction en Alberta. [Traduit par la Redaction] Mots-cles: gestion des interfaces, construction, rendement, analyse des facteurs, correlation., 1. Introduction Interface management (IM) is a key for the success of construction projects. Although there is no agreement among different previous researchers about the definition of IM, many discuss [...]

This article details the initial development of the School Counseling Program Implementation Survey and psychometric results including reliability and factor structure. An exploratory factor analysis revealed a three-factor model that [...]

The detailed geometry of atherosclerosis-prone vascular segments may influence their susceptibility by mediating local hemodynamics. An appreciation of the role of specific geometric variables is complicated by the considerable correlation among the many parameters that can be used to describe arterial shape and size. Factor analysis is a useful tool for identifying the essential features of such an inter-related data set, as well as for predicting hemodynamic risk in terms of these features and for interpreting the role of specific geometric variables. Here, factor analysis is applied to a set of 14 geometric variables obtained from magnetic resonance images of 50 human carotid bifurcations. Two factors alone were capable of predicting 12 hemodynamic metrics related to shear and near-wall residence time with adjusted squared Pearson's correlation coefficient as high as 0.54 and P-values less than 0.0001. One factor measures cross-sectional expansion at the bifurcation; the other measures the colinearity of the common and internal carotid artery axes at the bifurcation. The factors explain the apparent lack of an effect of branch angle on hemodynamic risk. The relative risk among the 50 bifurcations, based on time-average wall shear stress, could be predicted with a sensitivity and specificity as high as 0.84. The predictability of the hemodynamic metrics and relative risk is only modestly sensitive to assumptions about flow rates and flow partitions in the bifurcation. [DOI: 10.1115/1.4002538]

Exploring and exploiting variety by environment (V x E) interaction is one of the major challenges facing plant breeders. In paper I of this series, we presented an approach to modelling V x E interaction in the analysis of complex multi-environment trials using factor analytic models. In this paper, we develop a range of statistical tools which explore V x Einteraction in this context. These tools include graphical displays such as heat-maps of genetic correlation matrices as well as so-called E-scaled uniplots that are a more informative alternative to the classical biplot for large plant breeding multi-environment trials. We also present a new approach to prediction for multi-environment trials that include pedigree information. This approach allows meaningful selection indices to be formed either for potential new varieties or potential parents. Key words: multi-environment trials, factor analytic models, pedigrees, selection. L'exploration et l'exploitation des interactions genotype x environnement (G x E) est l'un des deefis les plus importants pour les selectionneurs. Dans l'article I de cette serie, les auteurs ont presente une approche pour modeliser les interactions G x E au moyen de modeles d'analyse des facteurs a partir des donnees d'essais complexes realises dans plusieurs environnements. Dans ce travail, les auteurs developpent une gamme d'outils statistiques qui permettent d'explorer les interactions G x E dans ce contexte. Ces outils incluent des outils de visualisation comme des cartes de densite des matrices de correlation geneetique ainsi que des graphiques appellees << E-scaled uniplots >> qui constituent une alternative plus informative aux classiques graphiques biplot employees pour les grands essais multi-environnements frequemment rencontresen selection. Les auteurs presentent egalement une nouvelle approche predictive pour les essais multi-environnements qui inclut l'information sur les pedigrees. Cette approche permet de former des indices de selection utiles pour la selection de nouveaux cultivars ou pour la selection de parents potentiels. Mots-cles: essais multi-environnements, modeles d'analyse des facteurs, pedigrees, selection. [Traduit par la Redaction], Introduction Beeck et al. (2010), paper I of this series, demonstrated the utility of the factor analytic (FA) model for capturing the variance structure of the variety by environment (V [...]

In this paper multiplicative mixed models have been used for the analysis of multi-environment trial (MET) data for canola oil and grain yield. Information on pedigrees has been included to allow for the modelling of additive and nonadditive genetic effects. The MET data set included a total of 19 trials (synonymous with sites or environments), which were sown across southern Australia in 2007 and 2008. Each trial was designed as a p-rep design using DiGGeR with the default prespecified spatial model. Lines in their first year of testing were unreplicated, whereas there were two or three replications of advanced lines or varieties. Pedigree information on a total of 578 entries was available, and there were 69 entries that had unknown pedigrees. The degree of inbreeding varied from 0 (55 entries) to nearly fully inbred (337 entries). Subsamples of 2 g harvested grain were taken from each plot for determination of seed oil percentage by near infrared reflectance spectroscopy. The MET analysis for both yield and oil modelled genetic effects in different trials using factor analytic models and the residual plot effects for each trial were modelled using spatial techniques. Models in which pedigree information was included provided significantly better fits to both yield and oil data. Key words: multi-environment trials, factor analytic models, pedigrees, spatial analysis. Dans ce travail, des modeles mixtes multiplicatifs ont ete employes pour l'analyse de donnees experimentales obtenues lors d'essais realises au champ dans plusieurs environnements pour mesurer le rendement en grains et en huile chez le canola. L'information sur les pedigrees a ete incluse pour permettre de modeliser les effets genetiques additifs et non-additifs. Le jeu de donnees comprenait un total de 19 essais (synonymes de sites ou d'environnements) semes dans le sud de l'Australie en 2007 et en 2008. Chaque essai a ete realise avec un dispositif p-rep a l'aide de DiGGeR en faisant appel au modele spatial specifie par defaut. Les lignees qui en etaient a leur premiere annee d'essai n'etaient pas repliquees, tandis que les lignees avancees et les varietes ont ete testees a raison de 2 a 3repetitions. Le pedigree de 578 lignees etait connu alors que celui de 69 lignees etait inconnu. Le degre de fixation genetique variait entre 0 (55 lignees) et presque completement fixe (337 lignees). Des echantillons de 2 g de grain ont ete preleves de chaque parcelle pour determiner la teneur en huile (%) par analyse infrarouge. L'analyse des donnees pour le rendement et l'huile a permis de modeliser les effets genetiques au sein des differents environnements a l'aide de modeles d'analyse des facteurs (FA) et les effets residuels des parcelles dans chaque environnement ont ete determinesa l'aide de techniques spatiales. Les modeles au sein desquels l'information concernant le pedigree etait incluse ont permis d'obtenir une meilleure concordance avec les donnees sur le rendement et la teneur en huile. Mots-cles: essais multi-environnements, modeles d'analyse des facteurs, pedigree, analyse spatiale. [Traduit par la Redaction], Introduction Mixed model approaches for analysing multi-environment trial (MET) data sets are in widespread use in many plant improvement programmes in Australia. These approaches are generally based on the models [...]

Problem statement: Impact of anthropogenic activities on man and his environment as a result of the growing rate of urbanization in Bida, Nigeria is of a great concern. Street sediments that accumulate along pavements in urban environments have the potential to provide considerable loadings of heavy metals to receiving waters and water bodies, particularly with changing environmental conditions. The objective of this research was to evaluate the streets sediment contamination in Bida, Nigeria. Approach: Fifty five sediment samples were collected from four roads that experience intense traffic conditions and analyzed in the laboratory for some heavy metals by atomic absorption Spectrophotometric method and multivariate statistical techniques. Results: The overall decreasing metal concentration order was: Pb > Mn > Fe > Zn > Cu > Cr > Ni > Cd. Significantly positive correlation was only found between Cd and organic matter (r = 0.580). Factor analysis shows that road deposited sediment quality data consists of four major components accounting for 77.11% of cumulative variance of the contamination: Ni, pH and silt + clay; Cr, Fe and organic matter; Mn and Zn and finally Cu and Pb. Discriminant analysis revealed that the first two Discriminate Functions (DF1 and DF2) contain 90.61% information for Cu, Pb and Ni accumulation. Conclusion: This study concluded that the concentrations of all metals measured in Bida can be considered to present a low level of contamination and that multivariate statistical analysis is a useful tool in understanding contaminants relationships. Key words: Anthropogenic, environment, multivariate, discriminant analysis, heavy metals, INTRODUCTION The presence of Metal pollutants in road-deposited sediments plays an important role in dictating urban storm water quality. Cities and Towns under various geographical, geological, climatic and sociological conditions [...]

In developing a classification model for assigning observations of unknown class to one of a number of specified classes using the values of a set of features associated with each observation, it is often desirable to base the classifier on a limited number of features. Mathematical programming discriminant analysis methods for developing classification models can be extended for feature selection. Classification accuracy can be used as the feature selection criterion by using a mixed integer programming (MIP) model in which a binary variable is associated with each training sample observation, but the binary variable requirements limit the size of problems to which this approach can be applied. Heuristic feature selection methods for problems with large numbers of observations are developed in this paper. These heuristic procedures, which are based on the MIP model for maximizing classification accuracy, are then applied to three credit scoring data sets. Journal of the Operational Research Society (2010) 61, 804-812. doi: 10.1057/jors.2009.24 Published online 8 April 2009 Keywords: discriminant analysis; mathematical programming; credit scoring

Raman spectroscopy is a label-free, real-time diagnostic tool that shows great promise in identifying cell differences. We have evaluated the discriminatory power of Raman spectroscopy using a unique model system consisting of two isogenic cancer cell lines derived from the MDA-MB-435 cell line. The two cell lines are equally tumorigenic in mice, but while M-4A4 gives rise to metastasis, NM-2C5 only disseminates single cells that remain dormant in distant organs. Previous comparative proteomic and transcriptomic analyses of the two cell lines have shown that they differ only in the expression level of a few proteins and genes. Raman maps were recorded of single cells after fixation and drying using 785 nm laser excitation. K-means clustering reduced the amount of data from each cell and improved the signal-to-noise ratio of cluster-averaged spectra. Spectra representing the nucleus were discarded as they showed much smaller differences between the two cell lines compared to cytoplasm spectra. Partial least squares-discriminant analysis (PLS-DA) was applied to distinguish the two cell lines. A cross-validated PLS-DA resulted in 92% correctly classified samples. Spectral differences were assigned to a higher unsaturated fatty acid content in the metastatic vs nonmetastatic cell line. Our study demonstrates the unique ability of Raman spectroscopy to distinguish minute differences at the subcellular level and yield new biological information. Our study is the first to demonstrate the association between polyunsaturated fatty acid content and metastatic ability in this unique cell model system and is in agreement with previous studies on this topic. 10.1021/ac902717d

Problem statement: The variations of formyl proton Chemical Shifts (CS) of p-substituted benzaldehydes in aromatic solvents were investigated. The validity of several physical solvent and empirical solvent scales was examined. Also, to predict dipolarity-polarizability ([pi]*) solvent scale for some aromatic solvents. Model designing was also achieved to rationalize the aromatic solvent effect on the formyl proton CS. Approach: The previously recorded formyl proton CS for p-X-benzaldehydes, with X were [NMe.sub.2], OMe, O[C.sub.3][H.sub.7], H, Br, CHO and [NO.sub.2] in benzene, toluene, p-xylene, m-xylene and mesitylene were subjected to Factor Analysis (FA). Target Factor Testing technique (TFT) was performed for several solvent scales namely: Unity, the intrinsic aromatic solvent induced shift of TMS (IASISTMS), f(n), f(d), ([n.sup.2]-1)/([n.sup.2]+2), (d-l)/(d+2), [E.sub.T](30) and [pi]*. Iterative TFT was applied to predict unmeasured ([pi]*) solvent scale for ethyl benzene, n-butyl benzene, sec-butyl benzene, tert-butyl benzene and isopropyl benzene. Results: It has been found that two factors were responsible for the variation in the formyl proton CS. The unity, f(n), ([n.sup.2]-l)/([n.sup.2]+2), IASISTMS, [E.sub.T](30) and [pi]* were real factors. Model designing of the formyl proton CS in benzene, toluene, p-xylene, m-xylene and mesitylene were achieved. The models with lowest root mean square error (RMSE) have shown that Unity is a consistent term. The other term was either IASISTMS or [pi]* . Iterative TFT predicted new [pi]* values for ethylbenzene, n-butylbenzene, sec-butylbenzene, tert-butylbenzene and isopropylbenzene respectively. Conclusion: FA has revealed that two real factors are responsible for the variation of formyl CS in benzene, toluene, p-xylene, m-xylene and mesitylene solvents. TFT has shown to be a powerful technique in predicting new values of the [pi]* solvent scale. Model designing for the formyl proton CS have revealed that the IASISTMS, [pi]* and Unity are the best empirical solvent scales and were better than any physical solvent scales in reproducing the formyl CS. The IASISTMS reflects the dipolarity-polarizabilty of the aromatic solvent. The cofactor of the solvent scale was found to correlate with the [[sigma].sub.p.sup.+] substituent parameter. Key words: Formyl proton NMR chemical shift, factor analysis, principal components analysis, target factor analysis, aromatic solvent effect, dipolarity-polarizability solvent scale, INTRODUCTION It is well-known that a solvent exerts an effect on many solvent dependent properties. These solvent-solute interactions can be related to physicochemical scales by constructing a solvent model (Koppel [...]

The impact of dependence between individual test statistics is currently among the most discussed topics in the multiple testing of high dimensional data literature, especially since Benjamini and Hoehberg (1995) introduced the false discovery rate (FDR). Many papers have first focused on the impact of dependence on the control of the FDR. Some more recent works have investigated approaches that account for common information shared by all the variables to stabilize the distribution of the error rates. Similarly, we propose to model this sharing of information by a factor analysis structure for the conditional variance of the test statistics. It is shown that the variance of the number of false discoveries increases along with the fraction of common variance. Test statistics for general linear contrasts are deduced, taking advantage of the common factor structure to reduce the variance of the error rates. A conditional FDR estimate is proposed and the overall performance of multiple testing procedure is shown to be markedly improved, regarding the nondiscovery rate, with respect to classical procedures. The present methodology is also assessed by comparison with leading multiple testing methods. KEY WORDS: Factor analysis; False discovery rate; Multiple-hypothesis testing; Nondiscovery rate.

This is a response to a recent paper in this journal, identifying a substantial error in the empirical example used to illustrate geographically weighted discriminant analysis and suggesting that it [...]

Introduction This is a short response to some recent queries relating to the example of geographically weighted discriminant analysis (GWDA) applied to U.K. election results in 2005 in a paper [...]

InfoMax and FastICA are the independent component analysis algorithms most used and apparently most effective for brain fMRI. We show that this is linked to their ability to handle effectively sparse components rather than independent components as such. The mathematical design of better analysis tools for brain fMRI should thus emphasize other mathematical characteristics than independence.

Traditional Western superstitious beliefs, such as black cats and the number 13 bringing bad luck, may not be applicable to different cultures. This study develops a Chinese Superstitious Belief Scale by conducting two studies with 363 and 395 participants, respectively. Exploratory factor analysis was used to construct the scale and then structural equation modeling was applied to verify its reliability and validity. The scale contains six dimensions, Homonym, Traditional customs, Power of crystal, Horoscope, Feng-shui, and Luck for gambling. Findings are helpful for understanding the difference between Chinese superstitions and the traditional Western superstitions and permits subsequent development of sociopsychological theories on correlates and effects of Chinese superstitions.

This study examined the construct validity of the motivational component of the Motivated Strategies for Learning Questionnaire using both exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). 327 students enrolled in Bachelor of Education and Graduate Diploma Education courses at the University of Auckland participated in this study (278 women, 49 men). The results of the EFA did not generate support for either a six-factor or three-factor model. The CFA tested a six-factor correlated original model and a respecified model and found a good fit for the respecified model and a better fit of the original 31-item model in comparison with previous studies. In addition, alternative second-order and hierarchical models were tested, none of which showed better fit than the six-factor correlated models. The reasons for the lack of fit of the original model are discussed in light of nonnormality condition, specification error, and latent variables. Finally, possible explanations for the slight variation in patterns of subscale correlations are discussed.

Kernel methods are a class of well established and successful algorithms for pattern analysis due to their mathematical elegance and good performance. Numerous nonlinear extensions of pattern recognition techniques have been proposed so far based on the so-called kernel trick. The objective of this paper is twofold. First, we derive an additional kernel tool that is still missing, namely kernel quadratic discriminant (KQD). We discuss different formulations of KQD based on the regularized kernel Mahalanobis distance in both complete and class-related subspaces. Second, we propose suitable extensions of kernel linear and quadratic discriminants to indefinite kernels. We provide classifiers that are applicable to kernels defined by any symmetric similarity measure. This is important in practice because problem-suited proximity measures often violate the requirement of positive definiteness. As in the traditional case, KQD can be advantageous for data with unequal class spreads in the kernel-induced spaces, which cannot be well separated by a linear discriminant. We illustrate this on artificial and real data for both positive definite and indefinite kernels. Index Terms--Machine learning, pattern recognition, kernel methods, indefinite kernels, discriminant analysis.

Forensic anthropology typically uses osteological and/or dental data either to estimate characteristics of unidentified individuals or to serve as evidence in cases where there is a putative identification. In the estimation context, the problem is to describe aspects of an individual that may lead to their eventual identification, whereas in the evidentiary context, the problem is to provide the relative support for the identification. In either context, individual characteristics such as sex and race may be useful. Using a previously published forensic case (Steadman et al. (2006) Am J Phys Anthropol 131:15-26) and a large (N = 3,167) reference sample, we show that the sex of the individual can be reliably estimated using a small set of 11 craniometric variables. The likelihood ratio from sex (assuming a 1:1 sex ratio for the 'population at large') is, however, relatively uninformative in 'making' the identification. Similarly, the known 'race' of the individual is relatively uninformative in 'making' the iden tification, because the individual was recovered from an area where the 2000 US census provides a very homogenous picture of (self-identified) race. Of interest in this analysis is the fact that the individual, who was recovered from Eastern Iowa, classifies very clearly with [Howells 1973. Cranial Variation in Man: A Study by Multivariate Analysis of Patterns of Difference Among Recent Human Populations. Cambridge, MA: Peabody Museum of Archaeology and Ethnology; 1989. Skull Shape and the Map: Craniometric Analyses in the Dispersion of Modern Homo. Cambridge, MA: Harvard University Press]. Easter Islander sample in an analysis with uninformative priors. When the Iowa 2000 Census data on self-reported race are used for informative priors, the individual is clearly identified as 'American White.' This analysis shows the extreme importance of an informative prior in any forensic application. KEY WORDS discriminant function; model-based clustering; likelihood ratio; craniometrics

An intelligent machine vision system is investigated and used for pattern recognition and classification of seven different types of cork tiles. The system includes image acquisition with a charge-coupled device (CCD) camera, texture feature generation (co-occurrence matrices and Laws' masks), analysis and processing of the feature vectors [linear discriminant analysis (LDA) and principal component analysis (PCA)], and cork tiles classification with feedforward neural networks (NN), employing our [GLP.sub.t]S (genetic low-discrepancy search) hybrid global optimization method. In addition, the same NN are trained with backpropagation (BP) and the obtained results are compared with the ones from [GLP.sub.[tau]S. The NN generalization abilities are discussed and assessed with respect to the NN architectures and the texture feature sets. The reported results are very encouraging with testing rate reaching up to 95 %. Index Terms--Image processing, linear discriminant analysis (LDA), machine vision, neural networks, optimization methods, pattern recognition, principal component analysis (PCA), texture feature extraction.

Abstract. -- A morphometric comparison was performed on specimens of blue catfish (Ictalurus furcatus) from northern (Lower Rio Bravo) and southern (Chiapas) Mexico in order to identify diagnostic characters that [...]

This paper analyzes the classification of hyperspectral remote sensing images with linear discriminant analysis (LDA) in the presence of a small ratio between the number of training samples and the number of spectral features. In these particular ill-posed problems, a reliable LDA requires one to introduce regularization for problem solving. Nonetheless, in such a challenging scenario, the resulting regularized LDA (RLDA) is highly sensitive to the tuning of the regularization parameter. In this context, we introduce in the remote sensing community an efficient version of the RLDA recently presented by Ye et al. to cope with critical ill-posed problems. In addition, several LDA-based classifiers (i.e., penalized LDA, orthogonal LDA, and uncorrelated LDA) are compared theoretically and experimentally with the standard LDA and the RLDA. Method differences are highlighted through toy examples and are exhaustively tested on several ill-posed problems related to the classification of hyperspectral remote sensing images. Experimental results confirm the effectiveness of the presented RLDA technique and point out the main properties of other analyzed LDA techniques in critical ill-posed hyperspectral image classification problems. Index Terms--Hyperspectral images, ill-posed problem, image classification, linear discriminant analysis (LDA), regularization.

In this paper, a new theoretical framework based on hidden Markov model (HMM) and independent component analysis (ICA) mixture model is presented for content analysis of video, namely ICAMHMM. Unlike the Gaussian mixture observation model commonly used in conventional HMM applications, the observations in the new ICAMHMM are modeled as a mixture of non-Gaussian components. Each non-Gaussian component is formulated by an ICA mixture, reflecting the independence of different components across video frames. In addition, to construct a compact feature space to represent a video frame, ICA is applied on video frames and the ICA coefficients are used to form a compact 2-D feature subspace that makes the subsequent modeling computationally efficient. The model parameters can be identified using supervised learning by the training sequences. The new re-estimation learning formulae of iterative ICAMHMM parameter estimation are derived based on a maximum likelihood function. Employing the identified model, maximum likelihood algorithms are developed to detect and recognize video events. As a case study, golf video sequences are used to test the effectiveness of the proposed algorithm. Experimental results show that the presented method can effectively detect and recognize the recurrent event patterns in video data. The presented new ICAMHMM is generic and can be applied to sequential data analysis in other applications. Index Terms--Hidden Markov model, independent component analysis (ICA) mixture model, sequential data analysis, video content analysis.

In this paper, a novel method for continuous human movement recognition based on fuzzy vector quantization (FVQ) and linear discriminant analysis (LDA) is proposed. We regard a movement as a unique combination of basic movement patterns, the so-called dynemes. The proposed algorithm combines FVQ and LDA to discover the most discriminative dynemes as well as represent and discriminate the different human movements in terms of these dynemes. This method allows for simple Mahalanobis or cosine distance comparison of not aligned human movements, taking into account implicitly time shifts and internal speed variations, and, thus, aiding the design of a real-time continuous human movement recognition algorithm. The effectiveness and robustness of this method is shown by experimental results on a standard dataset with videos captured under real conditions, and on a new video dataset created using motion capture data. Index Terms--Fuzzy vector quantization, linear discriminant analysis, real-time continuous human movement recognition.

The analysis of ECG signals is of fundamental importance for cardiac diagnosis. Conventional ECG recordings, however, use a limited number of channels (12) and each records a mixture of activities generated in different parts of the heart. Therefore, direct observation of the ECG signals collected on the body surface is likely an inefficient way to study and diagnose cardiac abnormalities. This study describes new experimental and analytical methods to capture more meaningful ECG component signals, each representing more directly a physical cardiac source. This study first describes a simply applied method for collecting high-density ECG signals. The recorded signals are then separated by independent component analysis (ICA) to obtain spatially fixed and temporally independent component activations. Results from five subjects show that P-, QRS-, and T-waves can be clearly separated from the recordings, suggesting ICA might be an effective and useful tool for high-density ECG analysis, interpretation, and diagnosis. Index Terms--Blind signal separation, ECG, high-density surface ECG, independent component analysis (ICA), noninvasive imaging.

In this brief, we propose a fast expectation conditional maximization (ECM) algorithm for maximum-likelihood (ML) estimation of mixtures of factor analyzers (MFA). Unlike the existing expectation--maximization (EM) algorithms such as the EM in Ghahramani and Hinton, 1996, and the alternating ECM (AECM) in McLachlan and Peel, 2003, where the missing data contains component--indicator vectors as well as latent factors, the missing data in our ECM consists of component--indicator vectors only. The novelty of our algorithm is that closed-form expressions in all conditional maximization (CM) steps are obtained explicitly, instead of resorting to numerical optimization methods. As revealed by experiments, the convergence of our ECM is substantially faster than EM and AECM regardless of whether assessed by central processing unit (CPU) time or number of iterations. Index Terms--Alternating expectation conditional maximization (AECM), expectation conditional maximization (ECM), expectation maximization (EM), maximum-likelihood estimation (MLE), mixture of factor analyzers (MFA).

In magnetoencephalography (MEG) and electroencephalography (EEG), independent component analysis is widely applied to separate brain signals from artifact components. A number of different methods have been proposed for the automatic or semiautomatic identification of artifact components. Most of the proposed methods are based on amplitude statistics of the decomposed MEG/EEG signal. We present a fully automated approach based on amplitude and phase statistics of decomposed MEG signals for the isolation of biological artifacts such as ocular, muscle, and cardiac artifacts (CAs). The performance of different artifact identification measures was investigated. In particular, we show that phase statistics is a robust and highly sensitive measure to identify strong and weak components that can be attributed to cardiac activity, whereas a combination of different measures is needed for the identification of artifacts caused by ocular and muscle activity. With the introduction of a rejection performance parameter, we are able to quantify the rejection quality for eye blinks and CAs. We demonstrate in a set of MEG data the good performance of the fully automated procedure for the removal of cardiac, ocular, and muscle artifacts. The new approach allows routine application to clinical measurements with small effect on the brain signal. Index Terms--Artifact reduction, blind source separation (BSS), independent component analysis (ICA), magnetoencephalography (MEG).

High-dimensional data are common in many domains, and dimensionality reduction is the key to cope with the curse-of-dimensionality. Linear discriminant analysis (LDA) is a well-known method for supervised dimensionality reduction. When dealing with high-dimensional and low sample size data, classical LDA suffers from the singularity problem. Over the years, many algorithms have been developed to overcome this problem, and they have been applied successfully in various applications. However, there is a lack of a systematic study of the commonalities and differences of these algorithms, as well as their intrinsic relationships. In this paper, a unified framework for generalized LDA is proposed, which elucidates the properties of various algorithms and their relationships. Based on the proposed framework, we show that the matrix computations involved in LDA-based algorithms can be simplified so that the cross-validation procedure for model selection can be performed efficiently. We conduct extensive experiments using a collection of high-dimensional data sets, including text documents, face images, gene expression data, and gene expression pattern images, to evaluate the proposed theories and algorithms. Index Terms--Dimensionality reduction, linear discriminant analysis (LDA), model selection, principal component analysis (PCA), regularization, visualization.

Factor analysis (FA) seeks to reveal the relationship between an observed vector variable and a latent variable of reduced dimension. It has been widely used in many applications involving high-dimensional data, such as image representation and face recognition. An intrinsic limitation of FA lies in its potentially poor performance when the data dimension is high, a problem known as curse of dimensionality. Motivated by the fact that images are inherently matrices, we develop, in this brief, an FA model for matrix-variate variables and present an efficient parameter estimation algorithm. Experiments on both toy and real-world image data demonstrate that the proposed matrix-variant FA model is more efficient and accurate than the classical FA approach, especially when the observed variable is high-dimensional and the samples available are limited. Index Terms--Conditional expectation maximization (EM), face recognition, factor analysis (FA), matrix.

A sensor array containing 9 cross-reactive sensing fluorescent elements with different affinity and selectivity to 10 metal cations ([Ca.sup.2+], [Mg.sup.2+], [Cd.sup.2+], [Hg.sup.2+], [Co.sup.2+], [Zn.sup.2+], [Cu.sup.2+], [Ni.sup.2+], [Al.sup.3+], [Ga.sup.3+]) is described. The discriminatory capacity of the array was tested at different ranges of pH and at different cation concentrations using linear discriminant analysis (LDA). Qualitative identification of cations can be determined with over 96% of accuracy in a concentration range covering 3 orders of a magnitude (5-5000 [micro]M). Quantitative analysis can be achieved with over 90% accuracy in the concentration range between 10 and 5000 [micro]M. The array performance was also tested in identification of nine different mineral water brands utilizing their various electrolyte compositions and their [Ca.sup.2+], [Mg.sup.2+], and [Zn.sup.2+] levels. LDA cross-validation routine shows 100% correct classification for all trials. Preliminary results suggest that similar arrays could be used in testing of the consistency of the purification and manufacturing process of purified and mineral waters.

In this study, the orthogonal projections to latent structures discriminant analysis (OPLS-DA) method was used to assess the in situ chemical composition of two different cell types in mouse liver samples, hepatocytes and erythrocytes. High spatial resolution FT-IR micro-spectroscopy equipped with a focal plan array (FPA) detector is capable of simultaneously recording over 4000 spectra from 64 x 64 pixels with a maximum spatial resolution of about 5 [micro]m x 5 [micro]m, which allows for the differentiation of individual cells. The main benefit with OPLS-DA lies in the ability to separate predictive variation (between cell type) from variation that is uncorrelated to cell type in order to facilitate understanding of different sources of variation. OPLS-DA was able to differentiate between chemical properties and physical properties (e.g., edge effects). OPLS-DA model interpretation of the chemical features that separated the two cell types clearly highlighted proteins and lipids/bile acids. The modeled variation that was uncorrelated to cell type made up a larger portion of the total variation and displayed strong variability in the amide I region. This could be traced back to a gradient in the high intensity (high-density) areas vs the low intensity areas (close to empty areas) that as a result of normalization had an adverse effect on FT-IR spectral profiles. This highlights that OPLS-DA provides an effective solution to identify different sources of variability, both predictive and uncorrelated, and also facilitates understanding of any sampling, experimental, or preprocessing issues.

We previously reported a method for the automated (objective) selection of a PARAFAC model having an appropriate number of factors for mathematical resolution of signal from a target analyte in GC x GC-TOFMS data (i.e., for an analysis in which the identity of the analyte is known a priori). While the previous target method has been successfully applied in several studies, the target method requires that the identity of the analyte be known. Also, multiple applications of the target method are required in cases where several analytes of interest are present in a single subsection of the chromatogram. Thus, having to know the analyte identity a priori restricts the applicability of the automated implementation of PARAFAC. The method presented in this report generalizes the previous method to allow analysis of one or more nontarget analyte signals in a subsection of a GC x GC-TOFMS chromatogram (i.e., for analyses when identities of analyte and interferences are not known a priori), thereby addressing and overcoming the limitations of the target method. Herein, we put the nontarget analyte PARAFAC method into theoretical context and illustrate the mechanics of the method using simulated data. We use real experimental GC x GC-TOFMS data to demonstrate the broad applicability of the method, with various analysis situations selected to illustrate challenging chemical analysis scenarios.

We address two shortcomings of the common spatial patterns (CSP) algorithm for spatial filtering in the context of brain--computer interfaces (BCIs) based on electroencephalography/magnetoencephalography (EEG/MEG): First, the question of optimality of CSP in terms of the minimal achievable classification error remains unsolved. Second, CSP has been initially proposed for two-class paradigms. Extensions to multiclass paradigms have been suggested, but are based on heuristics. We address these shortcomings in the framework of information theoretic feature extraction (ITFE). We show that for two-class paradigms, CSP maximizes an approximation of mutual information of extracted EEG/MEG components and class labels. This establishes a link between CSP and the minimal classification error. For multiclass paradigms, we point out that CSP by joint approximate diagonalization (JAD) is equivalent to independent component analysis (ICA), and provide a method to choose those independent components (ICs) that approximately maximize mutual information of ICs and class labels. This eliminates the need for heuristics in multiclass CSP, and allows incorporating prior class probabilities. The proposed method is applied to the dataset IIIa of the third BCI competition, and is shown to increase the mean classification accuracy by 23.4% in comparison to multiclass CSP. Index Terms--Brain-computer interfaces, common spatial patterns, information theoretic feature extraction, spatial filtering.

In this letter, we propose the application of the generalized eigenvalue decomposition for the decomposition of multichannel electrocardiogram (ECG) recordings. The proposed method uses a modified version of a previously presented measure of periodicity and a phase-wrapping of the RR-interval, for extracting the 'most periodic' linear mixtures of a recorded dataset. It is shown that the method is an improved extension of conventional source separation techniques, specifically customized for ECG signals. The method is therefore of special interest for the decomposition and compression of multichannel ECG, and for the removal of maternal ECG artifacts from fetal ECG recordings. Index Terms--Blind source separation, fetal ECG extraction, generalized eigenvalue decomposition.

The FastICA algorithm is one of the most prominent methods to solve the problem of linear independent component analysis (ICA). Although there have been several attempts to prove local convergence properties of FastICA, rigorous analysis is still missing in the community. The major difficulty of analysis is because of the well-known sign-flipping phenomenon of FastICA, which causes the discontinuity of the corresponding FastICA map on the unit sphere. In this paper, by using the concept of principal fiber bundles, FastICA is proven to be locally quadratically convergent to a correct separation. Higher order local convergence properties of FastICA are also investigated in the framework of a scalar shift strategy. Moreover, as a parallelized version of FastICA, the so-called QR FastICA algorithm, which employs the QR decomposition (Gram-Schmidt orthonormalization process) instead of the polar decomposition, is shown to share similar local convergence properties with the original FastICA. Index Terms--Approximate Newton method, convergence analysis, FastICA, independent component analysis (ICA), principal fiber bundle.

Fast independent component analysis (FastICA) algorithm separates the independent sources from their mixtures by measuring non-Gaussian. FastICA is a common offline method to identify artifact and interference from their mixtures such as electroencephalogram (EEG), magnetoencephalography (MEG), and electrocardiogram (ECG). Therefore, it is valuable to implement FastICA for real-time signal processing. In this paper, the FastICA algorithm is implemented in a field-programmable gate array (FPGA), with the ability of real-time sequential mixed signals processing by the proposed pipelined FastICA architecture. Moreover, in order to increase the numbers precision, the hardware floating-point (FP) arithmetic units had been carried out in the hardware FastICA. In addition, the proposed pipeline FastICA provides the high sampling rate (192 kHz) capability by hand coding the hardware FastICA in hardware description language (HDL). To verify the features of the proposed hardware FastICA, simulations are first performed, then real-time signal processing experimental results are presented using the fabricated platform. Experimental results demonstrate the effectiveness of the presented hardware FastICA as expected. Index Terms--Blind source separation (BSS), fast independent component analysis (FastICA), field-programmable gate array (FPGA), floating point (FP).

Independent component analysis (ICA) has found great promise in magnetic resonance (MR) image analysis. Unfortunately, two key issues have been overlooked and not investigated. One is the lack of MR images to be used to unmix signal sources of interest. Another is the use of random initial projection vectors by ICA, which causes inconsistent results. In order to address the first issue, this paper introduces a band-expansion process (BEP) to generate an additional new set of images from the original MR images via nonlinear functions. These newly generated images are then combined with the original MR images to provide sufficient MR images for ICA analysis. In order to resolve the second issue, a prioritized ICA (PICA) is designed to rank the ICA-generated independent components (ICs) so that MR brain tissue substances can be unmixed and separated by different ICs in a prioritized order. Finally, BEP and PICA are combined to further develop a new ICA-based approach, referred to as PICA-BEP to perform MR image analysis. Index Terms--Band-expansion process (BEP), FastICA, independent component analysis (ICA), magnetic resonance (MR) analysis, over-complete ICA (OC-ICA), prioritized ICA (PICA), prioritized ICA band-expansion process (PICA-BEP).

Here, we demonstrate that a single biochemical assay is able to predict the tissue-selective pharmacology of an array of selective estrogen receptor modulators (SERMs). We describe an approach to classify estrogen receptor (ER) modulators based on dynamics of the receptor-ligand complex as probed with hydrogen/deuterium exchange (HDX) mass spectrometry. Differential HDX mapping coupled with cluster and discriminate analysis effectively predicted tissue-selective function in most, but not all, cases tested. We demonstrate that analysis of dynamics of the receptor-ligand complex facilitates binning of ER modulators into distinct groups based on structural dynamics. Importantly, we were able to differentiate small structural changes within ER ligands of the same chemotype. In addition, HDX revealed differentially stabilized regions within the ligand-binding pocket that may contribute to the different pharmacology phenotypes of the compounds independent of helix 12 positioning. In summary, HDX provides a sensitive and rapid approach to classify modulators of the estrogen receptor that correlates with their pharmacological profile. discriminate analysis | hydrogen/deuterium exchange | mass spectrometry

This paper shows that if a measure of predictability is invariant to affine transformations and monotonically related to forecast uncertainty, then the component that maximizes this measure for normally distributed variables is independent of the detailed form of the measure. This result explains why different measures of predictability such as anomaly correlation, signal-to-noise ratio, predictive information, and the Mahalanobis error are each maximized by the same components. These components can be determined by applying principal component analysis to a transformed forecast ensemble, a procedure called predictable component analysis (PrCA). The resulting vectors define a complete set of components that can be ordered such that the first maximizes predictability, the second maximizes predictability subject to being uncorrelated of the first, and so on. The transformation in question, called the whitening transformation, can be interpreted as changing the norm in principal component analysis. The resulting norm renders noise variance analysis equivalent to signal variance analysis, whereas these two analyses lead to inconsistent results if other norms are chosen to define variance. Predictable components also can be determined by applying singular value decomposition to a whitened propagator in linear models. The whitening transformation is tantamount to changing the initial and final norms in the singular vector calculation. The norm for measuring forecast uncertainty has not appeared in prior predictability studies. Nevertheless, the norms that emerge from this framework have several attractive properties that make their use compelling. This framework generalizes singular vector methods to models with both stochastic forcing and initial condition error. These and other components of interest to predictability are illustrated with an empirical model for sea surface temperature.

High-dimensional data and the small sample size problem occur in many modern pattern classification applications such as face recognition and gene expression data analysis. To deal with such data, one important step is dimensionality reduction. Principal component analysis (PCA) and between-group analysis (BGA) are two commonly used methods, and various extensions of these two methods exist. The principle of these two approaches comes from their best approximation property. From a pattern recognition perspective, we show that PCA, which is based on the total scatter matrix, preserves linear separability, and BGA, which is based on between-class scatter matrix, retains the distance between class centroids. Moreover, we propose an automatic nonparameter uncorrelated discriminant analysis (UDA) algorithm based on the maximum margin criterion (MMC). The extracted features via UDA are statistically uncorrelated. UDA combines rank-preserving dimensionality reduction and constraint discriminant analysis and also serves as an effective solution for the small-sample-size problem. Experiments with face images and gene expression data sets are conducted to evaluate UDA in terms of classification accuracy and robustness. Index Terms--Feature extraction, feature interpretability, small sample size problem, face recognition, microarray data classification.

One of the most challenging problems for image retrieval applications is to find the optimal mapping between high-level semantic concept and low-level features. Traditional approaches often assume that images with same semantic label share strong visual similarities and should be clustered together to facilitate modeling and classification. Our research indicates this assumption is inappropriate in many cases. Instead we model the images as lying on nonlinear image subspaces embedded in the high-dimensional feature space and find that multiple subspaces may correspond to one semantic concept. By intelligently utilizing the similarity and dissimilarity information in semantic and geometric (image) domains, we propose an optimal semantic subspace projection (SSP) that captures the most important properties of the subspaces with respect to classification. Theoretical analysis proves that the well-known linear discriminant analysis (LDA) could be formulated as a special case of SSP. To capture the semantic concept dynamically, SSP can integrate relevance feedback efficiently through incremental learning. Kernel SSP is further proposed to handle nonlinearly separable data. Extensive experiments have been designed and conducted to compare our proposed method to the state-of-the-art techniques such as LDA, locality preservation projection (LPP), local linear embedding (LLE), local discriminant embedding (LDE) and their variants. The results show the superior performance of SSP. Index Terms--Image retrieval, linear discriminant analysis (LDA), principal component analysis, relevance feedback, semantic subspace projection (SSP), subspace learning.

The purposes of this study were to examine the reliability and validity of a 4-factor model of the Drinking Motives Measure and to assess year in school and ethnic differences on different types of motives. Data were collected on 441 volunteer college students. Results indicated that fit indices for the 4-factor model were acceptable; fit indices were also better for the 4-factor model than they were for alternative models. Freshman students and students of color had higher scores on the Conformity motives subscale than senior and White students did, respectively. Additionally, differences in the correlation between Confortuity motives and alcohol use existed based on year in school, such that the relationship was significantly smaller for freshmen than it was for other students. Keywords: college drinking, drinking motives, confirmatory factor analysis

In this paper, we use complex analytic functions to achieve independent component analysis (ICA) by maximization of non-Gaussianity and introduce the complex maximization of nonGaussianity (CMN) algorithm. We derive both a gradient-descent and a quasi-Newton algorithm that use the full second-order statistics providing superior performance with circular and noncircular sources as compared to existing methods. We show the connection among ICA methods through maximization of non-Gaussianity, mutual information, and maximum likelihood (ML) for the complex case, and emphasize the importance of density matching for all three cases. Local stability conditions are derived for the CMN cost function that explicitly show the effects of noncircularity on convergence and demonstrated through simulation examples. Index Terms--Complex-valued data, independent component analysis (ICA), quasi-Newton algorithm.

Large-margin methods, such as support vector machines (SVMs), have been very successful in classification problems. Recently, maximum margin discriminant analysis (MMDA) was proposed that extends the large-margin idea to feature extraction. It often outperforms traditional methods such as kernel principal component analysis (KPCA) and kernel Fisher discriminant analysis (KFD). Howvever, as in the SVM, its time complexity is cubic in the number of training points m, and is thus computationally inefficient on massive data sets. In this paper, we propose an [(1 + [epsilon]).sup.2]-approximation algorithm for obtaining the MMDA features by extending the core vector machine. The resultant time complexity is only linear in m, while its space complexity is independent of m. Extensive comparisons with the original MMDA, KPCA, and KFD on a number of large data sets show that the proposed feature extractor can improve classification accuracy, and is also faster than these kernel-based methods b v over an order of magnitude. Index Terms--Feature extraction, support vector machines (SVMs), core vector machines, scalability.

The determination of sex is one of the fundamental assessments in the production of a biological profile for human skeletal remains, and a large number of studies have focused on evaluating sexual dimorphism metrically. A recent study of a contemporary American population presented high classification accuracies from discriminant function analysis of the proximal ulna (Grant and Jant, 2003). The present research tests this method using a large sample (223 skeletons: 114 males and 109 females) from two skeletal assemblages of European origin, the documented Spitalfields Coffin-Plate Collection (N = 171) and the archaeological Raunds Furnells Collection (N = 52). The three measurements from the original study (Grant and Jantz, 2003), defined as the notch length (NL), the width of the olecranon process (OW), and the height of the coronoid process (CH) were tested, with the addition of a new variable radial notch height (RNH). Significant differences (P < 0.001) were found between male and female ulna measurements. Discriminant analysis found the original discriminant equation classified individuals with accuracies of only 40% for males but 99% for females. New discriminant functions were developed using direct and stepwise analyses of the Spitalfields sample. A direct multivariate function based on four dimensions (NL, OW, CH, RNH) produced an overall classification accuracy of 85.4% (males 82.4%; females 88.4%). A second function based on the original three measurements (NL, OW, CH) yielded nearly identical classification accuracies (males 81%; females 88.5%). Crossvalidation of the Spitalfields data using the three-variable function yielded an overall classification rate of 84.2%. This study indicates that the ulna is sexually dimorphic but its development is population-specific. KEY WORDS sexual dimorphism; discriminant function; Spitalfields; bioarchaeology; forensic anthropology