Learning shape priors for single view reconstruction

In this paper, we aim to reconstruct free-from 3D models from a single view by learning the prior knowledge of a specific class of objects. Instead of heuristically proposing specific regularities and defining parametric models as previous research, our shape prior is learned directly from existing 3D models under a framework based on the Gaussian Process Latent Variable Model (GPLVM). The major contributions of the paper include: 1) a probabilistic framework for prior-based reconstruction we propose, which requires no heuristic of the object, and can be easily generalized to handle various categories of 3D objects, and 2) an attempt at automatic reconstruction of more complex 3D shapes, like human bodies, from 2D silhouettes only. Qualitative and quantitative experimental results on both synthetic and real data demonstrate the efficacy of our new approach. ©2009 IEEE.

Warped Mixtures for Nonparametric Cluster Shapes

A mixture of Gaussians fit to a single curved or heavy-tailed cluster will report that the data contains many clusters. To produce more appropriate clusterings, we introduce a model which warps a latent mixture of Gaussians to produce nonparametric cluster shapes. The possibly low-dimensional latent mixture model allows us to summarize the properties of the high-dimensional clusters (or density manifolds) describing the data. The number of manifolds, as well as the shape and dimension of each manifold is automatically inferred. We derive a simple inference scheme for this model which analytically integrates out both the mixture parameters and the warping function. We show that our model is effective for density estimation, performs better than infinite Gaussian mixture models at recovering the true number of clusters, and produces interpretable summaries of high-dimensional datasets.

情绪干扰下焦虑水平对执行抑制的影响机制——事件相关电位研究

Behavioral inhibition model suggests the generation of anxiety is related with over-inhibition. For knowing about anxiety better, we used event-related potential (ERP) technique to explore the underlying mechanism of executive inhibition under the emotional distracter in high and low trait-anxious groups. Firstly, we set up the Chinese affective picture system (CAPS) as the stimuli of subsequent experiments. Secondly, we screened the high and low trait-anxious participants using the State-Trait Anxiety Inventory. In the first ERP study, a modified oddball paradigm was used with the positive, neutral and negative pictures as novel stimuli and the potentials evoked by three types pictures were analyzed. In the second ERP study, the same paradigm with higher task load was employed to examine the interaction of anxious level and emotion. Main results as follows: 1. CAPS consisted of 852 pictures was assessed via three dimensionalities, valence, arousal and dominance. The standard deviation of scores on valence and dominance was more than the standard deviation of scores on dimension of arousal. Scatter plot showed that the score distributing on the dimension of valence and arousal was wide in CAPS. 2. In both high and low trait-anxiety groups, the amplitudes of N2 and P3 of negative pictures were greater and smaller respectively as compared with neutral and positive pictures, which suggested all participants no matter what anxious level required more inhibition processing to negative information than others. 3. With increasing of task load, the P3 amplitudes of negative pictures in high anxious group were reduced relative to neutral pictures. In addition, in high anxious group, the P3 amplitudes of positive pictures had the same changes as those of negative ones. Whereas, the reduced P3 of positive pictures were not observed in low anxious group. The results showed the high anxious participants employed the same inhibitory strategy to the positive distracter as the negative distracter, which possibly the over-inhibition processing was involved in this group. 4. Dipole source analysis found cingulate may be involved in executive inhibition processing. In sum, as for the inhibition, high and low anxious group both is sensitive to negative information. However, in the high load situation, due to the shortness of cognitive resources, the high anxious individual represents the general sensitivity to all emotional information. These results gave the electrophysiological evidence for over-inhibition in high trait-anxiety group.

Tracking Human Body Pose on a Learned Smooth Space

Particle filtering is a popular method used in systems for tracking human body pose in video. One key difficulty in using particle filtering is caused by the curse of dimensionality: generally a very large number of particles is required to adequately approximate the underlying pose distribution in a high-dimensional state space. Although the number of degrees of freedom in the human body is quite large, in reality, the subset of allowable configurations in state space is generally restricted by human biomechanics, and the trajectories in this allowable subspace tend to be smooth. Therefore, a framework is proposed to learn a low-dimensional representation of the high-dimensional human poses state space. This mapping can be learned using a Gaussian Process Latent Variable Model (GPLVM) framework. One important advantage of the GPLVM framework is that both the mapping to, and mapping from the embedded space are smooth; this facilitates sampling in the low-dimensional space, and samples generated in the low-dimensional embedded space are easily mapped back into the original highdimensional space. Moreover, human body poses that are similar in the original space tend to be mapped close to each other in the embedded space; this property can be exploited when sampling in the embedded space. The proposed framework is tested in tracking 2D human body pose using a Scaled Prismatic Model. Experiments on real life video sequences demonstrate the strength of the approach. In comparison with the Multiple Hypothesis Tracking and the standard Condensation algorithm, the proposed algorithm is able to maintain tracking reliably throughout the long test sequences. It also handles singularity and self occlusion robustly.

Multiloop robust H∞ control design based on the dynamic PLS approach to chemical processes

In this paper, a multiloop robust control strategy is proposed based on H∞ control and a partial least squares (PLS) model (H∞_PLS) for multivariable chemical processes. It is developed especially for multivariable systems in ill-conditioned plants and non-square systems. The advantage of PLS is to extract the strongest relationship between the input and the output variables in the reduced space of the latent variable model rather than in the original space of the highly dimensional variables. Without conventional decouplers, the dynamic PLS framework automatically decomposes the MIMO process into multiple single-loop systems in the PLS subspace so that the controller design can be simplified. Since plant/model mismatch is almost inevitable in practical applications, to enhance the robustness of this control system, the controllers based on the H∞ mixed sensitivity problem are designed in the PLS latent subspace. The feasibility and the effectiveness of the proposed approach are illustrated by the simulation results of a distillation column and a mixing tank process. Comparisons between H∞_PLS control and conventional individual control (either H∞ control or PLS control only) are also made

The Determinants of Brownfields Redevelopment in England

This paper uses discrete choice models, supported by GIS data, to analyse the National Land Use Database, a register of more than 21,000 English brownfields - previously used sites with or without contamination that are currently unused or underused. Using spatial discrete choice models, including the first application of a spatial probit latent class model with class-specific neighbourhood effects, we find evidence of large local differences in the determinants of brownfields redevelopment in England and that the reuse decisions of adjacent sites affect the reuse of a site. We also find that sites with a history of industrial activities, large sites, and sites that are located in the poorest and bleakest areas of cities and regions of England are more difficult to redevelop. In particular, we find that the probability of reusing a brownfield increases by up to 8.5% for a site privately owned compared to a site publicly owned and between 15% - 30% if a site is located in London compared to the North West of England. We suggest that local tailored policies are more suitable than regional or national policies to boost the reuse of brownfield sites.

La salmonellose chez les bovins laitiers : présentation clinique et culture bactériologique

Huit cent trente et un troupeaux de vaches laitières répartis dans 5 états américains ont été enrôlés dans une étude de cohorte prospective. Un modèle d’équations d'estimation généralisées a été utilisé pour étudier l'association entre les signes cliniques et la détection de salmonelles dans les fèces des animaux soupçonnés de salmonellose clinique. La sensibilité et la spécificité de la culture bactériologique ont été estimées à l’aide d’un modèle de classes latentes. Dix-huit pour cent des 874 échantillons provenant de veaux et 29% des 1479 échantillons de vaches adultes étaient positifs pour Salmonella spp. Il n’a pas été possible d’établir une association claire entre les différents signes cliniques observés et la détection de salmonelles. Les 2 sérotypes les plus fréquemment isolés étaient Typhimurium et Newport. La probabilité de détecter des salmonelles était plus élevée chez les veaux où un autre agent entéropathogène était également détecté. La proportion d’échantillons positifs était plus élevée parmi les vaches ayant reçu des antibiotiques dans les jours précédant l’échantillonnage. La sensibilité de la culture a été estimée à 0,48 (intervalle de crédibilité à 95% [ICr95%]: 0,22-0,95) pour les veaux et 0,78 (ICr95%: 0,55-0,99) pour les vaches. La spécificité de la culture était de 0,94 (ICr95%: 0,87-1,00) pour les veaux et de 0,96 (ICr95%: 0,90-1,00) pour les vaches. Malgré une sensibilité imparfaite, la culture bactériologique demeure utile pour obtenir une meilleure estimation de la probabilité post-test de salmonellose clinique chez un bovin laitier, par rapport à la probabilité estimée suite au seul examen clinique.

Facteurs de risque modifiables associés à l’incidence, l’élimination et la prévalence d’infections intra-mammaires chez la vache laitière en lactation

La mammite subclinique est un problème de santé fréquent et coûteux. Les infections intra-mammaires (IIM) sont souvent détectées à l’aide de mesures du comptage des cellules somatiques (CCS). La culture bactériologique du lait est cependant requise afin d’identifier le pathogène en cause. À cause de cette difficulté, pratiquement toutes les recherches sur la mammite subclinique ont été centrées sur la prévalence d’IIM et les facteurs de risque pour l’incidence ou l’élimination des IIM sont peu connus. L’objectif principal de cette thèse était d’identifier les facteurs de risque modifiables associés à l’incidence, l’élimination et la prévalence d’IIM d’importance dans les troupeaux laitiers Canadiens. En premier lieu, une revue systématique de la littérature sur les associations entre pratiques utilisées à la ferme et CCS a été réalisée. Les pratiques de gestion constamment associées au CCS ont été identifiées et différentiées de celles faisant l’objet de rapports anecdotiques. Par la suite, un questionnaire bilingue a été développé, validé, et utilisé afin de mesurer les pratiques de gestion d’un échantillon de 90 troupeaux laitiers canadiens. Afin de valider l’outil, des mesures de répétabilité et de validité des items composant le questionnaire ont été analysées et une évaluation de l’équivalence des versions anglaise et française a été réalisée. Ces analyses ont permis d’identifier des items problématiques qui ont du être recatégorisés, lorsque possible, ou exclus des analyses subséquentes pour assurer une certaine qualité des données. La plupart des troupeaux étudiés utilisaient déjà la désinfection post-traite des trayons et le traitement universel des vaches au tarissement, mais beaucoup des pratiques recommandées n’étaient que peu utilisées. Ensuite, les facteurs de risque modifiables associés à l’incidence, à l’élimination et à la prévalence d’IIM à Staphylococcus aureus ont été investigués de manière longitudinale sur les 90 troupeaux sélectionnés. L’incidence d’IIM semblait être un déterminant plus important de la prévalence d’IIM du troupeau comparativement à l’élimination des IIM. Le port de gants durant la traite, la désinfection pré-traite des trayons, de même qu’une condition adéquate des bouts de trayons démontraient des associations désirables avec les différentes mesures d’IIM. Ces résultats viennent souligner l’importance des procédures de traite pour l’obtention d’une réduction à long-terme de la prévalence d’IIM. Finalement, les facteurs de risque modifiables associés à l’incidence, à l’élimination et à la prévalence d’IIM à staphylocoques coagulase-négatif (SCN) ont été étudiés de manière similaire. Cependant, afin de prendre en considération les limitations de la culture bactériologique du lait pour l’identification des IIM causées par ce groupe de pathogènes, une approche semi-Bayesienne à l’aide de modèles de variable à classe latente a été utilisée. Les estimés non-ajusté de l’incidence, de l’élimination, de la prévalence et des associations avec les expositions apparaissaient tous considérablement biaisés par les imperfections de la procédure diagnostique. Ce biais était en général vers la valeur nulle. Encore une fois, l’incidence d’IIM était le principal déterminant de la prévalence d’IIM des troupeaux. Les litières de sable et de produits du bois, de même que l’accès au pâturage étaient associés à une incidence et une prévalence plus basse de SCN.

Scene Classification Using a Hybrid Generative/Discriminative Approach

We investigate whether dimensionality reduction using a latent generative model is beneficial for the task of weakly supervised scene classification. In detail, we are given a set of labeled images of scenes (for example, coast, forest, city, river, etc.), and our objective is to classify a new image into one of these categories. Our approach consists of first discovering latent ";topics"; using probabilistic Latent Semantic Analysis (pLSA), a generative model from the statistical text literature here applied to a bag of visual words representation for each image, and subsequently, training a multiway classifier on the topic distribution vector for each image. We compare this approach to that of representing each image by a bag of visual words vector directly and training a multiway classifier on these vectors. To this end, we introduce a novel vocabulary using dense color SIFT descriptors and then investigate the classification performance under changes in the size of the visual vocabulary, the number of latent topics learned, and the type of discriminative classifier used (k-nearest neighbor or SVM). We achieve superior classification performance to recent publications that have used a bag of visual word representation, in all cases, using the authors' own data sets and testing protocols. We also investigate the gain in adding spatial information. We show applications to image retrieval with relevance feedback and to scene classification in videos

Image classification for a large number of object categories

L'increment de bases de dades que cada vegada contenen imatges més difícils i amb un nombre més elevat de categories, està forçant el desenvolupament de tècniques de representació d'imatges que siguin discriminatives quan es vol treballar amb múltiples classes i d'algorismes que siguin eficients en l'aprenentatge i classificació. Aquesta tesi explora el problema de classificar les imatges segons l'objecte que contenen quan es disposa d'un gran nombre de categories. Primerament s'investiga com un sistema híbrid format per un model generatiu i un model discriminatiu pot beneficiar la tasca de classificació d'imatges on el nivell d'anotació humà sigui mínim. Per aquesta tasca introduïm un nou vocabulari utilitzant una representació densa de descriptors color-SIFT, i desprès s'investiga com els diferents paràmetres afecten la classificació final. Tot seguit es proposa un mètode par tal d'incorporar informació espacial amb el sistema híbrid, mostrant que la informació de context es de gran ajuda per la classificació d'imatges. Desprès introduïm un nou descriptor de forma que representa la imatge segons la seva forma local i la seva forma espacial, tot junt amb un kernel que incorpora aquesta informació espacial en forma piramidal. La forma es representada per un vector compacte obtenint un descriptor molt adequat per ésser utilitzat amb algorismes d'aprenentatge amb kernels. Els experiments realitzats postren que aquesta informació de forma te uns resultats semblants (i a vegades millors) als descriptors basats en aparença. També s'investiga com diferents característiques es poden combinar per ésser utilitzades en la classificació d'imatges i es mostra com el descriptor de forma proposat juntament amb un descriptor d'aparença millora substancialment la classificació. Finalment es descriu un algoritme que detecta les regions d'interès automàticament durant l'entrenament i la classificació. Això proporciona un mètode per inhibir el fons de la imatge i afegeix invariança a la posició dels objectes dins les imatges. S'ensenya que la forma i l'aparença sobre aquesta regió d'interès i utilitzant els classificadors random forests millora la classificació i el temps computacional. Es comparen els postres resultats amb resultats de la literatura utilitzant les mateixes bases de dades que els autors Aixa com els mateixos protocols d'aprenentatge i classificació. Es veu com totes les innovacions introduïdes incrementen la classificació final de les imatges.

Gaussian processes autoencoder for dimensionality reduction

Learning low dimensional manifold from highly nonlinear data of high dimensionality has become increasingly important for discovering intrinsic representation that can be utilized for data visualization and preprocessing. The autoencoder is a powerful dimensionality reduction technique based on minimizing reconstruction error, and it has regained popularity because it has been efficiently used for greedy pretraining of deep neural networks. Compared to Neural Network (NN), the superiority of Gaussian Process (GP) has been shown in model inference, optimization and performance. GP has been successfully applied in nonlinear Dimensionality Reduction (DR) algorithms, such as Gaussian Process Latent Variable Model (GPLVM). In this paper we propose the Gaussian Processes Autoencoder Model (GPAM) for dimensionality reduction by extending the classic NN based autoencoder to GP based autoencoder. More interestingly, the novel model can also be viewed as back constrained GPLVM (BC-GPLVM) where the back constraint smooth function is represented by a GP. Experiments verify the performance of the newly proposed model.

Diethylpropion produces psychostimulant and reward effects

Diethylpropion (DEP) is a stimulant drug widely used for weight control in Brazil and other American countries. However, its effects on behavior and addiction potential are not yet well known. Data suggest that sensitization resulting from pre-exposure to psychostimulants could be a possible risk factor in subsequent drug addiction. The purpose of this investigation was to verify whether pre-exposure to DEP would sensitize rats to the motor activating effect and to the rewarding value of DEP. Two experiments were conducted. In both experiments rats were pre-exposed to DEP (20 mg/kg) or vehicle for 7 consecutive days. The acute effect of DEP (0.0, 1.0, 2.5 or 5.0 mg/kg) on motor activity (Experiment 1) and induction of Conditioned Place Preference-CPP (Experiment 2) were then measured. Results from Experiment 1 showed that 2.5 and 5.0 mg/kg DEP increased motor activity. Sensitization of this motor effect was observed. In Experiment 2, the doses of 2.5 and 5.0 mg/kg DEP induced CPP, indicating their rewarding value. However, no sensitization effect was observed. The results suggest that DEP at low doses has psychostimulant and rewarding properties. It is recommended that more effort should be dedicated to elucidating DEP abuse Potential. (c) 2008 Elsevier Inc. All rights reserved.

Performance of six diagnostic tests to screen for Chagas disease in blood banks and prevalence of Trypanosoma cruzi infection among donors with inconclusive serology screening based on the analysis of epidemiological variables

OBJECTIVE: The frequent occurrence of inconclusive serology in blood banks and the absence of a gold standard test for Chagas'disease led us to examine the efficacy of the blood culture test and five commercial tests (ELISA, IIF, HAI, c-ELISA, rec-ELISA) used in screening blood donors for Chagas disease, as well as to investigate the prevalence of Trypanosoma cruzi infection among donors with inconclusive serology screening in respect to some epidemiological variables. METHODS: To obtain estimates of interest we considered a Bayesian latent class model with inclusion of covariates from the logit link. RESULTS: A better performance was observed with some categories of epidemiological variables. In addition, all pairs of tests (excluding the blood culture test) presented as good alternatives for both screening (sensitivity > 99.96% in parallel testing) and for confirmation (specificity > 99.93% in serial testing) of Chagas disease. The prevalence of 13.30% observed in the stratum of donors with inconclusive serology, means that probably most of these are non-reactive serology. In addition, depending on the level of specific epidemiological variables, the absence of infection can be predicted with a probability of 100% in this group from the pairs of tests using parallel testing. CONCLUSION: The epidemiological variables can lead to improved test results and thus assist in the clarification of inconclusive serology screening results. Moreover, all combinations of pairs using the five commercial tests are good alternatives to confirm results.

Comparing Different Approaches for Clustering Categorical Data

There are different ways to do cluster analysis of categorical data in the literature and the choice among them is strongly related to the aim of the researcher, if we do not take into account time and economical constraints. Main approaches for clustering are usually distinguished into model-based and distance-based methods: the former assume that objects belonging to the same class are similar in the sense that their observed values come from the same probability distribution, whose parameters are unknown and need to be estimated; the latter evaluate distances among objects by a defined dissimilarity measure and, basing on it, allocate units to the closest group. In clustering, one may be interested in the classification of similar objects into groups, and one may be interested in finding observations that come from the same true homogeneous distribution. But do both of these aims lead to the same clustering? And how good are clustering methods designed to fulfil one of these aims in terms of the other? In order to answer, two approaches, namely a latent class model (mixture of multinomial distributions) and a partition around medoids one, are evaluated and compared by Adjusted Rand Index, Average Silhouette Width and Pearson-Gamma indexes in a fairly wide simulation study. Simulation outcomes are plotted in bi-dimensional graphs via Multidimensional Scaling; size of points is proportional to the number of points that overlap and different colours are used according to the cluster membership.

Analysis of Phenotypic Variation in Childhood Wheezing Disorders

Recurrent wheezing or asthma is a common problem in children that has increased considerably in prevalence in the past few decades. The causes and underlying mechanisms are poorly understood and it is thought that a numb er of distinct diseases causing similar symptoms are involved. Due to the lack of a biologically founded classification system, children are classified according to their observed disease related features (symptoms, signs, measurements) into phenotypes. The objectives of this PhD project were a) to develop tools for analysing phenotypic variation of a disease, and b) to examine phenotypic variability of wheezing among children by applying these tools to existing epidemiological data. A combination of graphical methods (multivariate co rrespondence analysis) and statistical models (latent variables models) was used. In a first phase, a model for discrete variability (latent class model) was applied to data on symptoms and measurements from an epidemiological study to identify distinct phenotypes of wheezing. In a second phase, the modelling framework was expanded to include continuous variability (e.g. along a severity gradient) and combinations of discrete and continuo us variability (factor models and factor mixture models). The third phase focused on validating the methods using simulation studies. The main body of this thesis consists of 5 articles (3 published, 1 submitted and 1 to be submitted) including applications, methodological contributions and a review. The main findings and contributions were: 1) The application of a latent class model to epidemiological data (symptoms and physiological measurements) yielded plausible pheno types of wheezing with distinguishing characteristics that have previously been used as phenotype defining characteristics. 2) A method was proposed for including responses to conditional questions (e.g. questions on severity or triggers of wheezing are asked only to children with wheeze) in multivariate modelling.ii 3) A panel of clinicians was set up to agree on a plausible model for wheezing diseases. The model can be used to generate datasets for testing the modelling approach. 4) A critical review of methods for defining and validating phenotypes of wheeze in children was conducted. 5) The simulation studies showed that a parsimonious parameterisation of the models is required to identify the true underlying structure of the data. The developed approach can deal with some challenges of real-life cohort data such as variables of mixed mode (continuous and categorical), missing data and conditional questions. If carefully applied, the approach can be used to identify whether the underlying phenotypic variation is discrete (classes), continuous (factors) or a combination of these. These methods could help improve precision of research into causes and mechanisms and contribute to the development of a new classification of wheezing disorders in children and other diseases which are difficult to classify.