Domain Adaptation in remote sensing: increasing the portability of land-cover classifiers

Among the types of remote sensing acquisitions, optical images are certainly one of the most widely relied upon data sources for Earth observation. They provide detailed measurements of the electromagnetic radiation reflected or emitted by each pixel in the scene. Through a process termed supervised land-cover classification, this allows to automatically yet accurately distinguish objects at the surface of our planet. In this respect, when producing a land-cover map of the surveyed area, the availability of training examples representative of each thematic class is crucial for the success of the classification procedure. However, in real applications, due to several constraints on the sample collection process, labeled pixels are usually scarce. When analyzing an image for which those key samples are unavailable, a viable solution consists in resorting to the ground truth data of other previously acquired images. This option is attractive but several factors such as atmospheric, ground and acquisition conditions can cause radiometric differences between the images, hindering therefore the transfer of knowledge from one image to another. The goal of this Thesis is to supply remote sensing image analysts with suitable processing techniques to ensure a robust portability of the classification models across different images. The ultimate purpose is to map the land-cover classes over large spatial and temporal extents with minimal ground information. To overcome, or simply quantify, the observed shifts in the statistical distribution of the spectra of the materials, we study four approaches issued from the field of machine learning. First, we propose a strategy to intelligently sample the image of interest to collect the labels only in correspondence of the most useful pixels. This iterative routine is based on a constant evaluation of the pertinence to the new image of the initial training data actually belonging to a different image. Second, an approach to reduce the radiometric differences among the images by projecting the respective pixels in a common new data space is presented. We analyze a kernel-based feature extraction framework suited for such problems, showing that, after this relative normalization, the cross-image generalization abilities of a classifier are highly increased. Third, we test a new data-driven measure of distance between probability distributions to assess the distortions caused by differences in the acquisition geometry affecting series of multi-angle images. Also, we gauge the portability of classification models through the sequences. In both exercises, the efficacy of classic physically- and statistically-based normalization methods is discussed. Finally, we explore a new family of approaches based on sparse representations of the samples to reciprocally convert the data space of two images. The projection function bridging the images allows a synthesis of new pixels with more similar characteristics ultimately facilitating the land-cover mapping across images.

Croire ensemble. Analyse institutionnelle du paysage religieux en Suisse

L'étude présentée dans cet ouvrage offre pour la première fois en Suisse une perspective sociologique empirique et générale sur les organisations religieuses locales. Vu sous cet angle, le fait religieux se manifeste d'abord, comme le précisaient déjà Weber ou Durkheim, par l'existence de communautés de tailles et de profils divers, tant dans l'espace rural qu'urbain. En Suisse, 5'734 paroisses et groupes religieux ont pu être dénombrés en 2008. Ce recensement permet de souligner la prégnance institutionnelle des Eglises historiques, mais également l'émergence d'une plus forte pluralité religieuse, particulièrement en milieu urbain. La sociologie des organisations que propose cette étude permet de jeter un éclairage nouveau sur les effets de la sécularisation (baisse des membres et de la pratique) et de la pluralisation (diversification des confessions et traditions religieuses). L'analyse des différences et des similitudes organisationnelles selon les traditions religieuses fait apparaître un positionnement social des groupes en fonction de statuts acquis au cours de l'histoire. Les théories classiques - par exemple à pro- pos des relations entre statuts sociaux des membres et groupes d'appartenance, des types d'autorités religieuses ou encore des différences entre Eglise et Secte - sont revisitées, précisées ou reformulées grâce à des données quantitatives originales et représentatives du champ religieux suisse. -- The study presented in this book for the first time in Switzerland offers a broad empirical and sociological perspective on local religious organizations. From this perspective, religion appears first, as already observed by Weber and Durkheim, through communities regardless of their various profiles. In Switzerland, 5,734 parishes and religious groups have been identified by a national census in 2008. The book underlines the institutional salience of historical churches (Reformed and Roman Catholic) but also the emerging religious plurality, particularly in urban areas. This study sheds a new light upon the effects of secularization (lower limbs and practice) and pluralisation (diversification of religious denominations and traditions) in the organizational field. The analysis of differences and similarities between religious traditions shows a social positioning of the local groups according to the statutes acquired in history. The classical theories − e.g. about the relationship between social status of members and membership groups, types of religious authorities or differences between Church and Sect − are revisited, clarified or reformulated based on an original and representative quantitative data of the Swiss religious field.

Ensemble methods for environmental data modelling with support vector regression

This paper investigates the use of ensemble of predictors in order to improve the performance of spatial prediction methods. Support vector regression (SVR), a popular method from the field of statistical machine learning, is used. Several instances of SVR are combined using different data sampling schemes (bagging and boosting). Bagging shows good performance, and proves to be more computationally efficient than training a single SVR model while reducing error. Boosting, however, does not improve results on this specific problem.

Apprendre ensemble pourtravailler ensemble: l'education interprofessionnelle, un mythe ou une réalité [Learning together to work together: interprofessional education, a myth or reality?].

This article offers a review of the literature on interprofessional education (EIP), a form of education which brings together members of two or more professions in a joint training. In this course, participants gain knowledge through other professionals and about them. The goal of EIP is to improve collaboration between health professionals and the quality of patient care. The EIP is booming worldwide and seems for from a mere fad. This expansion can be explained by several factors: the increasing importance attributed to the quality of care and patient safety, care changes (aging population and increasing chronic diseases) and the shortage of health professionals. The expectations of the EIP are large, while the evidence supporting its effectiveness is being built.

Transition aux soins adultes : une vue d'ensemble [Transition to adult care: an overview]

De nos jours, près de 90% des enfants atteints d'une maladie chronique survivent au-delà de l'âge de vingt ans et doivent passer de la pédiatrie aux soins adultes et de l'enfance à l'adolescence et à l'âge adulte. Selon la Society for Adolescent Medicine and Health (SAHM), les objectifs d'une transition organisée et coordonnée aux soins adultes pour les jeunes malades chroniques devraient permettre d'optimiser leur santé et de faciliter la réalisation de leur potentiel maximal.1 En conséquence, bien que le but principal de la transition soit la continuité des soins, elle n'est pas limitée au transfert (le passage de l'information et du patient de la pédiatrie aux soins adultes) mais est beaucoup plus large et inclut la préparation à la vie adulte. Ainsi donc, la transition devrait commencer tôt pendant l'adolescence, finir quand le patient devient un jeune adulte et englober trois parties : une phase de préparation en pédiatrie, une de transfert de la pédiatrie aux soins adultes et une dernière d'engagement aux soins adultes. Nowadays nearly 90% of children with a chronic condition survive to adulthood and must make the transition from pediatric to adult care. This transition must include not only the continuity of care but also the preparation for adult life so that these young people can develop their full potential. Divided into three phases (preparation, transfer and engagement), the transition process should be adapted to adolescents and ensure access to quality care.

Improved predictive mapping of indoor radon concentrations using ensemble regression trees based on automatic clustering of geological units.

PURPOSE: According to estimations around 230 people die as a result of radon exposure in Switzerland. This public health concern makes reliable indoor radon prediction and mapping methods necessary in order to improve risk communication to the public. The aim of this study was to develop an automated method to classify lithological units according to their radon characteristics and to develop mapping and predictive tools in order to improve local radon prediction. METHOD: About 240 000 indoor radon concentration (IRC) measurements in about 150 000 buildings were available for our analysis. The automated classification of lithological units was based on k-medoids clustering via pair-wise Kolmogorov distances between IRC distributions of lithological units. For IRC mapping and prediction we used random forests and Bayesian additive regression trees (BART). RESULTS: The automated classification groups lithological units well in terms of their IRC characteristics. Especially the IRC differences in metamorphic rocks like gneiss are well revealed by this method. The maps produced by random forests soundly represent the regional difference of IRCs in Switzerland and improve the spatial detail compared to existing approaches. We could explain 33% of the variations in IRC data with random forests. Additionally, the influence of a variable evaluated by random forests shows that building characteristics are less important predictors for IRCs than spatial/geological influences. BART could explain 29% of IRC variability and produced maps that indicate the prediction uncertainty. CONCLUSION: Ensemble regression trees are a powerful tool to model and understand the multidimensional influences on IRCs. Automatic clustering of lithological units complements this method by facilitating the interpretation of radon properties of rock types. This study provides an important element for radon risk communication. Future approaches should consider taking into account further variables like soil gas radon measurements as well as more detailed geological information.

The dynamics of interfaces: Seven authors in search of encounters across levels of description of an event involving a mother, father, and baby

The interfaces between the intrapsychic, interactional, and intergenerational domains are a new frontier. As a pilot, we exposed ourselves to a complex but controllable situation as viewed by people whose main interest is in one of the three interfaces; we also fully integrated the subjects in the team, to learn about their subjective perspectives and to provide them with an enriching experience. We started with a brief "triadification" sequence (i.e., moving from a "two plus one" to a "three together" family organization). Considering this sequence as representing at a micro level many larger family transitions, we proceeded with a microanalytic interview, a psychodynamic investigation, and a family interview. As expected, larger patterns of correspondences are emerging. Central questions under debate are: What are the most appropriate units at each level of description and what are their articulations between these levels? What is the status of "triadification"? Les interfaces entre les domaines intrapsychiques, interactionnels et intergénérationnels représentent une nouvelle frontiére. A titre exploratoire, nous nous sommes exposés à une situation complexe mais contrǒlable ainsi que le voient ceux dont I'intérět principal se porte sur l'une de ces trois interfaces. Nous avons aussi entièrement intégré les sujets dans l'équipe, de facon à comprendre leur perspective subjective et à leur offrir une expérience enrichissante. Nous avons commencé avec une brève séquence de "triadification," c'est-à-dire passer d'une organisation familiale "deux plus un" à Ltne organisation familiale "trois (add sentenc)ensemble." Considérant cette séquence comme representative à un niveau microscopique de transitions familiales bien plus larges, nous avons procedé à l'entretien microanalytique, à une enquěte psychodynamique et à un entretien familial. Comme prévu, de grands patterns de correspondances émergent. Les questions essentielles sur lesquelles portent le débat sont: quelles les unités les plus appropiées à chaque niveau de description et quelles sont les articulations entre ces niveaux? Quel est le statut de la "triadification"?