Advanced Kernel Methods For Remote Sensing Image Classification

Résumé Suite aux recentes avancées technologiques, les archives d'images digitales ont connu une croissance qualitative et quantitative sans précédent. Malgré les énormes possibilités qu'elles offrent, ces avancées posent de nouvelles questions quant au traitement des masses de données saisies. Cette question est à la base de cette Thèse: les problèmes de traitement d'information digitale à très haute résolution spatiale et/ou spectrale y sont considérés en recourant à des approches d'apprentissage statistique, les méthodes à noyau. Cette Thèse étudie des problèmes de classification d'images, c'est à dire de catégorisation de pixels en un nombre réduit de classes refletant les propriétés spectrales et contextuelles des objets qu'elles représentent. L'accent est mis sur l'efficience des algorithmes, ainsi que sur leur simplicité, de manière à augmenter leur potentiel d'implementation pour les utilisateurs. De plus, le défi de cette Thèse est de rester proche des problèmes concrets des utilisateurs d'images satellite sans pour autant perdre de vue l'intéret des méthodes proposées pour le milieu du machine learning dont elles sont issues. En ce sens, ce travail joue la carte de la transdisciplinarité en maintenant un lien fort entre les deux sciences dans tous les développements proposés. Quatre modèles sont proposés: le premier répond au problème de la haute dimensionalité et de la redondance des données par un modèle optimisant les performances en classification en s'adaptant aux particularités de l'image. Ceci est rendu possible par un système de ranking des variables (les bandes) qui est optimisé en même temps que le modèle de base: ce faisant, seules les variables importantes pour résoudre le problème sont utilisées par le classifieur. Le manque d'information étiquétée et l'incertitude quant à sa pertinence pour le problème sont à la source des deux modèles suivants, basés respectivement sur l'apprentissage actif et les méthodes semi-supervisées: le premier permet d'améliorer la qualité d'un ensemble d'entraînement par interaction directe entre l'utilisateur et la machine, alors que le deuxième utilise les pixels non étiquetés pour améliorer la description des données disponibles et la robustesse du modèle. Enfin, le dernier modèle proposé considère la question plus théorique de la structure entre les outputs: l'intègration de cette source d'information, jusqu'à présent jamais considérée en télédétection, ouvre des nouveaux défis de recherche. Advanced kernel methods for remote sensing image classification Devis Tuia Institut de Géomatique et d'Analyse du Risque September 2009 Abstract The technical developments in recent years have brought the quantity and quality of digital information to an unprecedented level, as enormous archives of satellite images are available to the users. However, even if these advances open more and more possibilities in the use of digital imagery, they also rise several problems of storage and treatment. The latter is considered in this Thesis: the processing of very high spatial and spectral resolution images is treated with approaches based on data-driven algorithms relying on kernel methods. In particular, the problem of image classification, i.e. the categorization of the image's pixels into a reduced number of classes reflecting spectral and contextual properties, is studied through the different models presented. The accent is put on algorithmic efficiency and the simplicity of the approaches proposed, to avoid too complex models that would not be used by users. The major challenge of the Thesis is to remain close to concrete remote sensing problems, without losing the methodological interest from the machine learning viewpoint: in this sense, this work aims at building a bridge between the machine learning and remote sensing communities and all the models proposed have been developed keeping in mind the need for such a synergy. Four models are proposed: first, an adaptive model learning the relevant image features has been proposed to solve the problem of high dimensionality and collinearity of the image features. This model provides automatically an accurate classifier and a ranking of the relevance of the single features. The scarcity and unreliability of labeled. information were the common root of the second and third models proposed: when confronted to such problems, the user can either construct the labeled set iteratively by direct interaction with the machine or use the unlabeled data to increase robustness and quality of the description of data. Both solutions have been explored resulting into two methodological contributions, based respectively on active learning and semisupervised learning. Finally, the more theoretical issue of structured outputs has been considered in the last model, which, by integrating outputs similarity into a model, opens new challenges and opportunities for remote sensing image processing.

Role of glast-positive glial cells during photoreceptor development

Purpose: Taking advantage of two transgenic lines, glast.DsRed and crx.gfp, that express fluorescent proteins in glial and photoreceptor cells respectively, we investigate the role of glast-positive glial cells (GPCs) in the survival/differentiation/proliferation of age-matched photoreceptor cells. Methods: Primary retinal cells were isolated from newborn transgenic mouse retina (glast.dsRed::crx.gfp) at postnatal day (P0/P1) and propagated in defined medium containing epidermal growth factor (EGF) and fibroblast growth factor 2 (bFGF). By flow-sorting another population of pure GPCs was isolated. Both populations were expanded and analyzed for the presence of specific retinal cell markers. Notably, the primary cell culture collected from the transgenic line glast.dsRed::crx.gfp showed a conspicuous presence of immature photoreceptors growing on top of GPCs. In order to reveal the role of such cells in the survival/differentiation/proliferation of photoreceptors we set up in vitro cultures of retina-derived cells that allowed long-term time-lapse recordings charting every cell division, death and differentiation event. To assess the regenerative potential of GPCs we challenged them with compounds mimicking retinal degeneration (NMU, NMDA, Zaprinast). Mass spectrometry (MS), immunostainings and other molecular approaches were performed to reveal adhesion molecules involved in the relationship between glial cells and photoreceptors. Results: Both primary cell lines were highly homogenous, with an elongated morphology and the majority expressed Müller glia markers (MG) such as glast, blbp, glt-1, vimentin, glutamine synthetase (GS), GFAP, cd44, mash1 and markers of reactive Müller glia such as nestin, pax6. Conversely, none of them were found positive for retinal neuron markers like tuj1, otx2, recoverin. Primary cultures of GPCs show the incapability of glial cells to give rise to photoreceptors in both wild type or degenerative environment. Furthermore, primary cultures of pure GPCs challenged with different compounds did not highlight the production of new glial cell-derived photoreceptors. Adhesion molecules involved in the contact between photoreceptors and glial cells are still under investigation. Conclusions: Primary glia cells do not give rise to photoreceptor cells in wt and degenerative conditions at least in vitro. The roles of glial cells seem to be more linked to the maintenance/proliferation of photoreceptor cells.

Negative and positive effects of trade on health.

Comment on: Blouin C, Chopra M, van der Hoeven R.Trade and social determinants of health. Lancet. 2009;373(9662):502-7. PMID: 19167058.

DEA does not like positive discrimination : a comparison of alternative models based on empirical data

Due to the existence of free software and pedagogical guides, the use of data envelopment analysis (DEA) has been further democratized in recent years. Nowadays, it is quite usual for practitioners and decision makers with no or little knowledge in operational research to run themselves their own efficiency analysis. Within DEA, several alternative models allow for an environment adjustment. Five alternative models, each of them easily accessible to and achievable by practitioners and decision makers, are performed using the empirical case of the 90 primary schools of the State of Geneva, Switzerland. As the State of Geneva practices an upstream positive discrimination policy towards schools, this empirical case is particularly appropriate for an environment adjustment. The alternative of the majority of DEA models deliver divergent results. It is a matter of concern for applied researchers and a matter of confusion for practitioners and decision makers. From a political standpoint, these diverging results could lead to potentially opposite decisions. Grâce à l'existence de logiciels en libre accès et de guides pédagogiques, la méthode data envelopment analysis (DEA) s'est démocratisée ces dernières années. Aujourd'hui, il n'est pas rare que les décideurs avec peu ou pas de connaissances en recherche opérationnelle réalisent eux-mêmes leur propre analyse d'efficience. A l'intérieur de la méthode DEA, plusieurs modèles permettent de tenir compte des conditions plus ou moins favorables de l'environnement. Cinq de ces modèles, facilement accessibles et applicables par les décideurs, sont utilisés pour mesurer l'efficience des 90 écoles primaires du canton de Genève, Suisse. Le canton de Genève pratiquant une politique de discrimination positive envers les écoles défavorisées, ce cas pratique est particulièrement adapté pour un ajustement à l'environnement. La majorité des modèles DEA génèrent des résultats divergents. Ce constat est préoccupant pour les chercheurs appliqués et perturbant pour les décideurs. D'un point de vue politique, ces résultats divergents conduisent à des prises de décision différentes selon le modèle sur lequel elles sont fondées.

The effect of continuous positive airway pressure on total cerebral blood flow in healthy awake volunteers.

PURPOSE: Continuous positive airway pressure (CPAP) is the gold standard treatment for obstructive sleep apnea. However, the physiologic impact of CPAP on cerebral blood flow (CBF) is not well established. Ultrasound can be used to estimate CBF, but there is no widespread accepted protocol. We studied the physiologic influence of CPAP on CBF using a method integrating arterial diameter and flow velocity (FV) measurements obtained for each vessel supplying blood to the brain. METHODS: FV and lumen diameter of the left and right internal carotid, vertebral, and middle cerebral arteries were measured using duplex Doppler ultrasound with and without CPAP at 15 cm H(2)O, applied in a random order. Transcutaneous carbon dioxide (PtcCO(2)), heart rate (HR), blood pressure (BP), and oxygen saturation were monitored. Results were compared with a theoretical prediction of CBF change based on the effect of partial pressure of carbon dioxide on CBF. RESULTS: Data were obtained from 23 healthy volunteers (mean ± SD; 12 male, age 25.1 ± 2.6 years, body mass index 21.8 ± 2.0 kg/m(2)). The mean experimental and theoretical CBF decrease under CPAP was 12.5 % (p < 0.001) and 11.9 % (p < 0.001), respectively. The difference between experimental and theoretical CBF reduction was not statistically significant (3.84 ± 79 ml/min, p = 0.40). There was a significant reduction in PtcCO(2) with CPAP (p = <0.001) and a significant increase in mean BP (p = 0.0017). No significant change was observed in SaO(2) (p = 0.21) and HR (p = 0.62). CONCLUSION: Duplex Doppler ultrasound measurements of arterial diameter and FV allow for a noninvasive bedside estimation of CBF. CPAP at 15 cm H(2)O significantly decreased CBF in healthy awake volunteers. This effect appeared to be mediated predominately through the hypocapnic vasoconstriction coinciding with PCO(2) level reduction. The results suggest that CPAP should be used cautiously in patients with unstable cerebral hemodynamics.

Three-dimensional black-blood cardiac magnetic resonance coronary vessel wall imaging detects positive arterial remodeling in patients with nonsignificant coronary artery disease.

BACKGROUND: Direct noninvasive visualization of the coronary vessel wall may enhance risk stratification by quantifying subclinical coronary atherosclerotic plaque burden. We sought to evaluate high-resolution black-blood 3D cardiovascular magnetic resonance (CMR) imaging for in vivo visualization of the proximal coronary artery vessel wall. METHODS AND RESULTS: Twelve adult subjects, including 6 clinically healthy subjects and 6 patients with nonsignificant coronary artery disease (10% to 50% x-ray angiographic diameter reduction) were studied with the use of a commercial 1.5 Tesla CMR scanner. Free-breathing 3D coronary vessel wall imaging was performed along the major axis of the right coronary artery with isotropic spatial resolution (1.0x1.0x1.0 mm(3)) with the use of a black-blood spiral image acquisition. The proximal vessel wall thickness and luminal diameter were objectively determined with an automated edge detection tool. The 3D CMR vessel wall scans allowed for visualization of the contiguous proximal right coronary artery in all subjects. Both mean vessel wall thickness (1.7+/-0.3 versus 1.0+/-0.2 mm) and wall area (25.4+/-6.9 versus 11.5+/-5.2 mm(2)) were significantly increased in the patients compared with the healthy subjects (both P<0.01). The lumen diameter (3.6+/-0.7 versus 3.4+/-0.5 mm, P=0.47) and lumen area (8.9+/-3.4 versus 7.9+/-3.5 mm(2), P=0.47) were similar in both groups. CONCLUSIONS: Free-breathing 3D black-blood coronary CMR with isotropic resolution identified an increased coronary vessel wall thickness with preservation of lumen size in patients with nonsignificant coronary artery disease, consistent with a "Glagov-type" outward arterial remodeling. This novel approach has the potential to quantify subclinical disease.

The aggregate claims distribution of a life insurance portfolio with a pairwise positive dependence structure

The individual life model has always been considered as the one closest to the real situation of the total claims of a life insurance portfolio. It only makes the ¿nearly inevitable assumption¿ of independence of the lifelenghts of insured persons in the portfolio. Many clinical studies, however, have demonstrated positive dependence of paired lives such as husband and wife. In our opinion, it won¿t be unrealistic expecting a considerable number of married couples in any life insurance portfolio (e.g. life insurance contracts formalized at the time of signing a mortatge) and these dependences materially increase the values for the stop-loss premiums associated to the aggregate claims of the portfolio. Since the stop-loss order is the order followed by any risk averse decison maker, the simplifying hypothesis of independence constitute a real financial danger for the company, in the sense that most of their decisions are based on the aggregated claims distribution. In this paper, we will determine approximations for the distribution of the aggregate claims of a life insurance portfolio with some married couples and we will describe how to make safe decisions when we don¿t know exactly the dependence structure between the risks in each couple. Results in this paper are partly based on results in Dhaene and Goovaerts (1997)

The Lorenz-maximal core allocations and the kernel in some classes of assignment games

En aquest treball demostrem que en la classe de jocs d'assignació amb diagonal dominant (Solymosi i Raghavan, 2001), el repartiment de Thompson (que coincideix amb el valor tau) és l'únic punt del core que és maximal respecte de la relació de dominància de Lorenz, i a més coincideix amb la solucié de Dutta i Ray (1989), també coneguda com solució igualitària. En segon lloc, mitjançant una condició més forta que la de diagonal dominant, introduïm una nova classe de jocs d'assignació on cada agent obté amb la seva parella òptima almenys el doble que amb qualsevol altra parella. Per aquests jocs d'assignació amb diagonal 2-dominant, el repartiment de Thompson és l'únic punt del kernel, i per tant el nucleolo.

The impact of fasting on the interpretation of triglyceride levels for predicting myocardial infarction risk in HIV-positive individuals: the D:A:D study.

We assessed whether fasting modifies the prognostic value of these measurements for the risk of myocardial infarction (MI). Analyses used mixed effect models and Poisson regression. After confounders were controlled for, fasting triglyceride levels were, on average, 0.122 mmol/L lower than nonfasting levels. Each 2-fold increase in the latest triglyceride level was associated with a 38% increase in MI risk (relative rate, 1.38; 95% confidence interval, 1.26-1.51); fasting status did not modify this association. Our results suggest that it may not be necessary to restrict analyses to fasting measurements when considering MI risk.

Mechanical ventilation in the newborn; a simplified approach. Part 1: Intermittent positive pressure ventilation.

Positive pressure ventilation (PPV) is a frequent intervention in the neonatal intensive care unit. This article is directed towards paediatricians in training and attempts to cover the basics of PPV without being too technical. To do so we have employed an extensive use of graphics to illustrate the underlying principles.

Direct in vitro comparison of six three-dimensional positive contrast methods for susceptibility marker imaging.

PURPOSE: To compare different techniques for positive contrast imaging of susceptibility markers with MRI for three-dimensional visualization. As several different techniques have been reported, the choice of the suitable method depends on its properties with regard to the amount of positive contrast and the desired background suppression, as well as other imaging constraints needed for a specific application. MATERIALS AND METHODS: Six different positive contrast techniques are investigated for their ability to image at 3 Tesla a single susceptibility marker in vitro. The white marker method (WM), susceptibility gradient mapping (SGM), inversion recovery with on-resonant water suppression (IRON), frequency selective excitation (FSX), fast low flip-angle positive contrast SSFP (FLAPS), and iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL) were implemented and investigated. RESULTS: The different methods were compared with respect to the volume of positive contrast, the product of volume and signal intensity, imaging time, and the level of background suppression. Quantitative results are provided, and strengths and weaknesses of the different approaches are discussed. CONCLUSION: The appropriate choice of positive contrast imaging technique depends on the desired level of background suppression, acquisition speed, and robustness against artifacts, for which in vitro comparative data are now available.

Multiple Kernel Learning of Environmental Data. Case study: Analysis and Mapping of Wind Fields

The paper presents the Multiple Kernel Learning (MKL) approach as a modelling and data exploratory tool and applies it to the problem of wind speed mapping. Support Vector Regression (SVR) is used to predict spatial variations of the mean wind speed from terrain features (slopes, terrain curvature, directional derivatives) generated at different spatial scales. Multiple Kernel Learning is applied to learn kernels for individual features and thematic feature subsets, both in the context of feature selection and optimal parameters determination. An empirical study on real-life data confirms the usefulness of MKL as a tool that enhances the interpretability of data-driven models.

The structure of nervous tissue and of Gram-positive bacteria studied by cryo-electron microscopy of vitreous sections

Résumé La structure, ou l'architecture, des êtres vivants définit le cadre dans lequel la physique de la vie s'accomplit. La connaissance de cette structure dans ses moindres détails est un but essentiel de la biologie. Son étude est toutefois entravée par des limitations techniques. Malgré son potentiel théorique, la microscopie électronique n'atteint pas une résolution atomique lorsqu'elle est appliquée ä la matièxe biologique. Cela est dû en grande partie au fait qu'elle contient beaucoup d'eau qui ne résiste pas au vide du microscope. Elle doit donc être déshydratée avant d'être introduite dans un microscope conventionnel. Des artéfacts d'agrégation en découlent inévitablement. La cryo-microscopie électronique des sections vitreuses (CEMOVIS) a ëté développée afin de résoudre cela. Les spécimens sont vitrifiés, c.-à-d. que leur eau est immobilisée sans cristalliser par le froid. Ils sont ensuite coupés en sections ultrafines et celles-ci sont observées à basse température. Les spécimens sont donc observés sous forme hydratée et non fixée; ils sont proches de leur état natif. Durant longtemps, CEMOVIS était très difficile à exécuter mais ce n'est plus le cas. Durant cette thèse, CEMOVIS a été appliqué à différents spécimens. La synapse du système nerveux central a été étudiée. La présence dans la fente synaptique d'une forte densité de molécules organisées de manière périodique a été démontrée. Des particules luminales ont été trouvées dans Ies microtubules cérébraux. Les microtubules ont servi d'objets-test et ont permis de démontrer que des détails moléculaires de l'ordre du nm sont préservés. La compréhension de la structure de l'enveloppe cellulaire des bactéries Grampositives aété améliorée. Nos observations ont abouti à l'élaboration d'un nouveau modèle hypothétique de la synthèse de la paroi. Nous avons aussi focalisé notre attention sur le nucléoïde bactérien et cela a suscité un modèle de la fonction des différents états structuraux du nucléoïde. En conclusion, cette thèse a démontré que CEMOVIS est une excellente méthode poux étudier la structure d'échantillons biologiques à haute résolution. L'étude de la structure de divers aspects des êtres vivants a évoqué des hypothèses quant à la compréhension de leur fonctionnement. Summary The structure, or the architecture, of living beings defines the framework in which the physics of life takes place. Understanding it in its finest details is an essential goal of biology. Its study is however hampered by technical limitations. Despite its theoretical potential, electron microscopy cannot resolve individual atoms in biological matter. This is in great part due to the fact. that it contains a lot of water that cannot stand the vacuum of the microscope. It must therefore be dehydrated before being introduced in a conventional mìcroscope. Aggregation artefacts unavoidably happen. Cryo-electron microscopy of vitreous sections (CEMOVIS) has been developed to solve this problem. Specimens are vitrified, i.e. they are rapidly cooled and their water is immobilised without crystallising by the cold. They are then. sectioned in ultrathin slices, which are observed at low temperatures. Specimens are therefore observed in hydrated and unfixed form; they are close to their native state. For a long time, CEMOVIS was extremely tedious but this is not the case anymore. During this thesis, CEMOVIS was applied to different specimens. Synapse of central nervous system was studied. A high density of periodically-organised molecules was shown in the synaptic cleft. Luminal particles were found in brain microtubules. Microtubules, used as test specimen, permitted to demonstrate that molecular details of the order of nm .are preserved. The understanding of the structure of cell envelope of Gram-positive bacteria was improved. Our observations led to the elaboration of a new hypothetic model of cell wall synthesis. We also focused our attention on bacterial nucleoids and this also gave rise to a functional model of nucleoid structural states. In conclusion, this thesis demonstrated that CEMOVIS is an excellent method for studying the structure of bìologìcal specimens at high resolution. The study of the structure of various aspects of living beings evoked hypothesis for their functioning.