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.

Lack of functionally active Melan-A(26-35)-specific T cells in the blood of HLA-A2+ vitiligo patients.

Vitiligo, a skin disorder characterized by the spontaneous destruction of melanocytes, is believed to be of autoimmune origin. We investigated the presence and functionality of CD8(+) T-cells specific for the melanocyte-associated antigens Melan-A, gp100, tyrosinase, and TRP-2 in the blood of HLA-A2(+) vitiligo patients. We enumerated antigen-specific CD8(+) T cells by major histocompatibility complex multimer staining directly ex vivo, as well as after 9 days of in vitro stimulation and assessed IFN-gamma secretion by enzyme-linked immunospot (Elispot) assay. Tyrosinase-, gp100-, or TRP-2-specific CD8(+) T cells could not be identified in the peripheral blood of individuals with vitiligo. Although Melan-A-specific T cells were detectable at levels comparable to Flu-MP-specific T cells by multimer staining, these lymphocytes did not express the skin-homing receptor cutaneous lymphocyte antigen, were phenotypically naïve (CD45RA(+)), and were unresponsive in the IFN-gamma Elispot assay, suggesting that they are unlikely to be involved in the etiopathogenesis of vitiligo.

Novel Macrocyclic Amidinoureas: Potent Non-Azole Antifungals Active against Wild-Type and Resistant Candida Species.

Novel macrocyclic amidinourea derivatives 11, 18, and 25 were synthesized and evaluated as antifungal agents against wild-type and fluconazole resistant Candida species. Macrocyclic compounds 11 and 18 were synthesized through a convergent approach using as a key step a ring-closing metathesis macrocyclization reaction, whereas compounds 25 were obtained by our previously reported synthetic pathway. All the macrocyclic amidinoureas showed antifungal activity toward different Candida species higher or comparable to fluconazole and resulted highly active against fluconazole resistant Candida strains showing in many cases minimum inhibitory concentration values lower than voriconazole.

Activation of the NLRP3 inflammasome by Mycobacterium tuberculosis is uncoupled from susceptibility to active tuberculosis.

As a hallmark of tuberculosis (TB), Mycobacterium tuberculosis (MTB) induces granulomatous lung lesions and systemic inflammatory responses during active disease. Molecular regulation of inflammation is associated with inflammasome assembly. We determined the extent to which MTB triggers inflammasome activation and how this impacts on the severity of TB in a mouse model. MTB stimulated release of mature IL-1β in macrophages while attenuated M. bovis BCG failed to do so. Tubercle bacilli specifically activated the NLRP3 inflammasome and this propensity was strictly controlled by the virulence-associated RD1 locus of MTB. However, Nlrp3-deficient mice controlled pulmonary TB, a feature correlated with NLRP3-independent production of IL-1β in infected lungs. Our studies demonstrate that MTB activates the NLRP3 inflammasome in macrophages in an ESX-1-dependent manner. However, during TB, MTB promotes NLRP3- and caspase-1-independent IL-1β release in myeloid cells recruited to lung parenchyma and thus overcomes NLRP3 deficiency in vivo in experimental models.

Short-term overexpression of a constitutively active form of AMP-activated protein kinase in the liver leads to mild hypoglycemia and fatty liver.

AMP-activated protein kinase (AMPK) is a major therapeutic target for the treatment of diabetes. We investigated the effect of a short-term overexpression of AMPK specifically in the liver by adenovirus-mediated transfer of a gene encoding a constitutively active form of AMPKalpha2 (AMPKalpha2-CA). Hepatic AMPKalpha2-CA expression significantly decreased blood glucose levels and gluconeogenic gene expression. Hepatic expression of AMPKalpha2-CA in streptozotocin-induced and ob/ob diabetic mice abolished hyperglycemia and decreased gluconeogenic gene expression. In normal mouse liver, AMPKalpha2-CA considerably decreased the refeeding-induced transcriptional activation of genes encoding proteins involved in glycolysis and lipogenesis and their upstream regulators, SREBP-1 (sterol regulatory element-binding protein-1) and ChREBP (carbohydrate response element-binding protein). This resulted in decreases in hepatic glycogen synthesis and circulating lipid levels. Surprisingly, despite the inhibition of hepatic lipogenesis, expression of AMPKalpha2-CA led to fatty liver due to the accumulation of lipids released from adipose tissue. The relative scarcity of glucose due to AMPKalpha2-CA expression led to an increase in hepatic fatty acid oxidation and ketone bodies production as an alternative source of energy for peripheral tissues. Thus, short-term AMPK activation in the liver reduces blood glucose levels and results in a switch from glucose to fatty acid utilization to supply energy needs.

The Political Economy of Active Labor-Market Policy

Active labor-market policies (ALMPs) have developed significantly over the past two decades across Organization for Economic Cooperation and Development (OECD) countries, with substantial cross-national differences in terms of both extent and overall orientation. The objective of this article is to account for cross-national variation in this policy field. It starts by reviewing existing scholarship concerning political, institutional, and ideational determinants of ALMPs. It then argues that ALMP is too broad a category to be used without further specification, and it develops a typology of four different types of ALMPs: incentive reinforcement, employment assistance, occupation, and human capital investment. These are discussed and examined through ALMP expenditure profiles in selected countries. The article uses this typology to analyze ALMP trajectories in six Western European countries and shows that the role of this instrument changes dramatically over time. It concludes that there is little regularity in the political determinants of ALMPs. In contrast, it finds strong institutional and ideational effects, nested in the interaction between the changing economic context and existing labor-market policies.

Organisational reforms in active welfare states: A comparative analysis of the turn to 'single gateways' in Western Europe

Based on the case of reforms aimed at integrating the provision of income protection and employment services for jobless people in Europe, this thesis seeks to understand the reasons which may prompt governments to engage in large-scale organisational reforms. Over the last 20 years, several European countries have indeed radically redesigned the organisational structure of their welfare state by merging or bundling existing front-line offices in charge of benefit payment and employment services together into 'one-stop' agencies. Whereas in academic and political debates, these reforms are generally presented as a necessary and rational response to the problems and inconsistencies induced by fragmentation in a context of the reorientation of welfare states towards labour market activation, this thesis shows that the agenda setting of these reforms is in fact the result of multidimensional political dynamics. More specifically, the main argument of this thesis is that these reforms are best understood not so such from the problems induced by organisational compartmentalism, whose political recognition is often controversial, but from the various goals that governments may simultaneously achieve by means of their adoption. This argument is tested by comparing agenda-setting processes of large-scale reforms of coordination in the United Kingdom (Jobcentre Plus), Germany (Hartz IV reform) and Denmark (2005 Jobcentre reform), and contrasting them with the Swiss case where the government has so far rejected any coordination initiative involving organisational redesign. This comparison brings to light the importance, for the rise of organisational reforms, of the possibility to couple them with the following three goals: first, goals related to the strengthening of activation policies; second, institutional goals seeking to redefine the balance of responsibilities between the central state and non-state actors, and finally electoral goals for governments eager to maintain political credibility. The decisive role of electoral goals in the three countries suggests that these reforms are less bound by partisan politics than by the particular pressures facing governments arrived in office after long periods in opposition.

Regular wave propagation out of noise in chemical active media

A pacemaker, regularly emitting chemical waves, is created out of noise when an excitable photosensitive Belousov-Zhabotinsky medium, strictly unable to autonomously initiate autowaves, is forced with a spatiotemporal patterned random illumination. These experimental observations are also reproduced numerically by using a set of reaction-diffusion equations for an activator-inhibitor model, and further analytically interpreted in terms of genuine coupling effects arising from parametric fluctuations. Within the same framework we also address situations of noise-sustained propagation in subexcitable media.

Effects of SCH 34826, an orally active inhibitor of atrial natriuretic peptide degradation, in healthy volunteers.

Atrial natriuretic peptide is cleared from plasma by clearance receptors and by enzymatic degradation by way of a neutral metalloendopeptidase. Inhibition of neutral metalloendopeptidase activity appears to provide an interesting approach to interfere with metabolism of atrial natriuretic peptide to enhance the renal and haemodynamic effects of endogenous atrial natriuretic peptide. In this study, the effects of SCH 34826, a new orally active neutral metalloendopeptidase inhibitor, have been evaluated in a single-blind, placebo-controlled study involving eight healthy volunteers who had maintained a high sodium intake for 5 days. SCH 34826 had no effect on blood pressure or heart rate in these normotensive subjects. SCH 34826 promoted significant increases in excretion of urinary sodium, phosphate, and calcium. The cumulative 5-hour urinary sodium excretion was 15.7 +/- 7.3 mmol for the placebo and 22.9 +/- 5, 26.7 +/- 6 (p less than 0.05), and 30.9 +/- 6.8 mmol (p less than 0.01) for the 400, 800, and 1600 mg SCH 34826 doses, respectively. During the same time interval, the cumulative urinary phosphate excretion increased by 0.3 +/- 0.4 mmol after placebo and by 1.5 +/- 0.3 (p less than 0.01), 1.95 +/- 0.3 (p less than 0.01), and 2.4 +/- 0.4 mmol (p less than 0.001) after 400, 800, and 1600 mg SCH 34826, respectively. There was no change in diuresis or excretion of urinary potassium and uric acid. The natriuretic response to SCH 34826 occurred in the absence of any change in plasma atrial natriuretic peptide levels but was associated with a dose-dependent elevation of urinary atrial natriuretic peptide and cyclic guanosine monophosphate.(ABSTRACT TRUNCATED AT 250 WORDS)