Extreme Precipitation Modelling Using Geostatistics and Machine Learning Algorithms

The paper presents an approach for mapping of precipitation data. The main goal is to perform spatial predictions and simulations of precipitation fields using geostatistical methods (ordinary kriging, kriging with external drift) as well as machine learning algorithms (neural networks). More practically, the objective is to reproduce simultaneously both the spatial patterns and the extreme values. This objective is best reached by models integrating geostatistics and machine learning algorithms. To demonstrate how such models work, two case studies have been considered: first, a 2-day accumulation of heavy precipitation and second, a 6-day accumulation of extreme orographic precipitation. The first example is used to compare the performance of two optimization algorithms (conjugate gradients and Levenberg-Marquardt) of a neural network for the reproduction of extreme values. Hybrid models, which combine geostatistical and machine learning algorithms, are also treated in this context. The second dataset is used to analyze the contribution of radar Doppler imagery when used as external drift or as input in the models (kriging with external drift and neural networks). Model assessment is carried out by comparing independent validation errors as well as analyzing data patterns.

Activity in inferior parietal and medial prefrontal cortex signals the accumulation of evidence in a probability learning task.

In an uncertain environment, probabilities are key to predicting future events and making adaptive choices. However, little is known about how humans learn such probabilities and where and how they are encoded in the brain, especially when they concern more than two outcomes. During functional magnetic resonance imaging (fMRI), young adults learned the probabilities of uncertain stimuli through repetitive sampling. Stimuli represented payoffs and participants had to predict their occurrence to maximize their earnings. Choices indicated loss and risk aversion but unbiased estimation of probabilities. BOLD response in medial prefrontal cortex and angular gyri increased linearly with the probability of the currently observed stimulus, untainted by its value. Connectivity analyses during rest and task revealed that these regions belonged to the default mode network. The activation of past outcomes in memory is evoked as a possible mechanism to explain the engagement of the default mode network in probability learning. A BOLD response relating to value was detected only at decision time, mainly in striatum. It is concluded that activity in inferior parietal and medial prefrontal cortex reflects the amount of evidence accumulated in favor of competing and uncertain outcomes.

Dynamics of the recruitment of lateral amygdala neurons following successive fear learning to different conditional stimuli

In fear conditioning, an animal learns to associate an unconditioned stimulus (US), such as a shock, and a conditioned stimulus (CS), such as a tone, so that the presentation of the CS alone can trigger conditioned responses. Recent research on the lateral amygdala has shown that following cued fear conditioning, only a subset of higher-excitable neurons are recruited in the memory trace. Their selective deletion after fear conditioning results in a selective erasure of the fearful memory. I hypothesize that the recruitment of highly excitable neurons depends on responsiveness to stimuli, intrinsic excitability and local connectivity. In addition, I hypothesize that neurons recruited for an initial memory also participate in subsequent memories, and that changes in neuronal excitability affect secondary fear learning. To address these hypotheses, I will show that A) a rat can learn to associate two successive short-term fearful memories; B) neuronal populations in the LA are competitively recruited in the memory traces depending on individual neuronal advantages, as well as advantages granted by the local network. By performing two successive cued fear conditioning experiments, I found that rats were able to learn and extinguish the two successive short-term memories, when tested 1 hour after learning for each memory. These rats were equipped with a system of stable extracellular recordings that I developed, which allowed to monitor neuronal activity during fear learning. 233 individual putative pyramidal neurons could modulate their firing rate in response to the conditioned tone (conditioned neurons) and/or non- conditioned tones (generalizing neurons). Out of these recorded putative pyramidal neurons 86 (37%) neurons were conditioned to one or both tones. More precisely, one population of neurons encoded for a shared memory while another group of neurons likely encoded the memories' new features. Notably, in spite of a successful behavioral extinction, the firing rate of those conditioned neurons in response to the conditioned tone remained unchanged throughout memory testing. Furthermore, by analyzing the pre-conditioning characteristics of the conditioned neurons, I determined that it was possible to predict neuronal recruitment based on three factors: 1) initial sensitivity to auditory inputs, with tone-sensitive neurons being more easily recruited than tone- insensitive neurons; 2) baseline excitability levels, with more highly excitable neurons being more likely to become conditioned; and 3) the number of afferent connections received from local neurons, with neurons destined to become conditioned receiving more connections than non-conditioned neurons. - En conditionnement de la peur, un animal apprend à associer un stimulus inconditionnel (SI), tel un choc électrique, et un stimulus conditionné (SC), comme un son, de sorte que la présentation du SC seul suffit pour déclencher des réflexes conditionnés. Des recherches récentes sur l'amygdale latérale (AL) ont montré que, suite au conditionnement à la peur, seul un sous-ensemble de neurones plus excitables sont recrutés pour constituer la trace mnésique. Pour apprendre à associer deux sons au même SI, je fais l'hypothèse que les neurones entrent en compétition afin d'être sélectionnés lors du recrutement pour coder la trace mnésique. Ce recrutement dépendrait d'un part à une activation facilité des neurones ainsi qu'une activation facilité de réseaux de neurones locaux. En outre, je fais l'hypothèse que l'activation de ces réseaux de l'AL, en soi, est suffisante pour induire une mémoire effrayante. Pour répondre à ces hypothèses, je vais montrer que A) selon un processus de mémoire à court terme, un rat peut apprendre à associer deux mémoires effrayantes apprises successivement; B) des populations neuronales dans l'AL sont compétitivement recrutées dans les traces mnésiques en fonction des avantages neuronaux individuels, ainsi que les avantages consentis par le réseau local. En effectuant deux expériences successives de conditionnement à la peur, des rats étaient capables d'apprendre, ainsi que de subir un processus d'extinction, pour les deux souvenirs effrayants. La mesure de l'efficacité du conditionnement à la peur a été effectuée 1 heure après l'apprentissage pour chaque souvenir. Ces rats ont été équipés d'un système d'enregistrements extracellulaires stables que j'ai développé, ce qui a permis de suivre l'activité neuronale pendant l'apprentissage de la peur. 233 neurones pyramidaux individuels pouvaient moduler leur taux d'activité en réponse au son conditionné (neurones conditionnés) et/ou au son non conditionné (neurones généralisant). Sur les 233 neurones pyramidaux putatifs enregistrés 86 (37%) d'entre eux ont été conditionnés à un ou deux tons. Plus précisément, une population de neurones code conjointement pour un souvenir partagé, alors qu'un groupe de neurones différent code pour de nouvelles caractéristiques de nouveaux souvenirs. En particulier, en dépit d'une extinction du comportement réussie, le taux de décharge de ces neurones conditionné en réponse à la tonalité conditionnée est resté inchangée tout au long de la mesure d'apprentissage. En outre, en analysant les caractéristiques de pré-conditionnement des neurones conditionnés, j'ai déterminé qu'il était possible de prévoir le recrutement neuronal basé sur trois facteurs : 1) la sensibilité initiale aux entrées auditives, avec les neurones sensibles aux sons étant plus facilement recrutés que les neurones ne répondant pas aux stimuli auditifs; 2) les niveaux d'excitabilité des neurones, avec les neurones plus facilement excitables étant plus susceptibles d'être conditionnés au son ; et 3) le nombre de connexions reçues, puisque les neurones conditionné reçoivent plus de connexions que les neurones non-conditionnés. Enfin, nous avons constaté qu'il était possible de remplacer de façon satisfaisante le SI lors d'un conditionnement à la peur par des injections bilatérales de bicuculline, un antagoniste des récepteurs de l'acide y-Aminobutirique.

Active Learning of Very-High Resolution Optical Imagery with SVM: Entropy vs Margin Sampling

An active learning method is proposed for the semi-automatic selection of training sets in remote sensing image classification. The method adds iteratively to the current training set the unlabeled pixels for which the prediction of an ensemble of classifiers based on bagged training sets show maximum entropy. This way, the algorithm selects the pixels that are the most uncertain and that will improve the model if added in the training set. The user is asked to label such pixels at each iteration. Experiments using support vector machines (SVM) on an 8 classes QuickBird image show the excellent performances of the methods, that equals accuracies of both a model trained with ten times more pixels and a model whose training set has been built using a state-of-the-art SVM specific active learning method

Machine learning algorithms for spatial data. Case studies: Environmental pollution, natural hazards, renewable resources

This paper presents general problems and approaches for the spatial data analysis using machine learning algorithms. Machine learning is a very powerful approach to adaptive data analysis, modelling and visualisation. The key feature of the machine learning algorithms is that they learn from empirical data and can be used in cases when the modelled environmental phenomena are hidden, nonlinear, noisy and highly variable in space and in time. Most of the machines learning algorithms are universal and adaptive modelling tools developed to solve basic problems of learning from data: classification/pattern recognition, regression/mapping and probability density modelling. In the present report some of the widely used machine learning algorithms, namely artificial neural networks (ANN) of different architectures and Support Vector Machines (SVM), are adapted to the problems of the analysis and modelling of geo-spatial data. Machine learning algorithms have an important advantage over traditional models of spatial statistics when problems are considered in a high dimensional geo-feature spaces, when the dimension of space exceeds 5. Such features are usually generated, for example, from digital elevation models, remote sensing images, etc. An important extension of models concerns considering of real space constrains like geomorphology, networks, and other natural structures. Recent developments in semi-supervised learning can improve modelling of environmental phenomena taking into account on geo-manifolds. An important part of the study deals with the analysis of relevant variables and models' inputs. This problem is approached by using different feature selection/feature extraction nonlinear tools. To demonstrate the application of machine learning algorithms several interesting case studies are considered: digital soil mapping using SVM, automatic mapping of soil and water system pollution using ANN; natural hazards risk analysis (avalanches, landslides), assessments of renewable resources (wind fields) with SVM and ANN models, etc. The dimensionality of spaces considered varies from 2 to more than 30. Figures 1, 2, 3 demonstrate some results of the studies and their outputs. Finally, the results of environmental mapping are discussed and compared with traditional models of geostatistics.

A New Twist on Radiation Oncology: Low-Dose Irradiation Elicits Immunostimulatory Macrophages that Unlock Barriers to Tumor Immunotherapy.

Tumor-infiltrating macrophages typically promote angiogenesis while suppressing antitumoral T cell responses. In this issue of Cancer Cell, Klug and colleagues report that clinically-feasible, low-dose irradiation redirects macrophage differentiation from a tumor-promoting/immunosuppressive state to one that enables cytotoxic T cells to infiltrate tumors and kill cancer cells, rendering immunotherapy successful in mice.

Characterization of root apoplastic barriers in Arabidopsis thaliana

In vascular plants, the best-known feature of a differentiated endodermal cell is the "Casparian Strip" (CS). This structure refers to a highly localized cell wall impregnation in the transversal and anticlinal walls of the cell, which surrounds the cell like a belt/ring and is tightly coordinated with respect to neighboring cells. Analogous to tight junctions in animal epithelia, CS in plants act as a diffusion barrier that controls the movement of water and ions from soil into the stele. Since its first description by Robert Caspary in 1865 there have been many attempts to identify the chemical nature of the cell wall deposition in CS. Suberin, lignin, or both have been claimed to be the important components of CS in a series of different species. However, the exact chemical composition of CS has remained enigmatic. This controversy was due to the confusion and lack of knowledge regarding the precise measurement of three developmental stages of the endodermis. The CS represent only the primary stage of endodermal differentiation, which is followed by the deposition of suberin lamellae all around the cellular surface of endodermal cells (secondary developmental stage). Therefore, chemical analysis of whole roots, or even of isolated endodermal tissues, will always find both of the polymers present. It was crucial to clarify this point because this will guide our efforts to understand which cell wall biosynthetic component becomes localized in order to form the CS. The main aim of my work was to find out the major components of (early) CS, as well as their spatial and temporal development, physiological roles and relationship to barrier formation. Employing the knowledge and tools that have been accumulated over the last few years in the model plant Arabidopsis thaliana, various histological and chemical assays were used in this study. A particular feature of my work was to completely degrade, or inhibit formation of lignin and suberin biopolymers by biochemical, classical genetic and molecular approaches and to investigate its effect on CS formation and the establishment of a functional diffusion barrier. Strikingly, interference with monolignol biosynthesis abrogates CS formation and delays the formation of function diffusion barrier. In contrast, transgenic plants devoid of any detectable suberin still develop a functional CS. The combination of all these assays clearly demonstrates that the early CS polymer is made from monolignol (lignin monomers) and is composed of lignin. By contrast, suberin is formed much later as a secondary wall during development of endodermis. These early CS are functionally sufficient to block extracellular diffusion and suberin does not play important role in the establishment of early endodermal diffusion barrier. Moreover, suberin biosynthetic machinery is not present at the time of CS formation. Our study finally concludes the long-standing debate about the chemical nature of CS and opens the door to a new approach in lignin research, specifically for the identification of the components of the CS biosynthetic pathway that mediates the localized deposition of cell walls. I also made some efforts to understand the patterning and differentiation of endodermal passage cells in young roots. In the literature, passage cells are defined as a non- suberized xylem pole associated endodermal cells. Since these cells only contain the CS but not the suberin lamellae, it has been assumed that these cells may offer a continued low-resistance pathway for water and minerals into the stele. Thus far, no genes have been found to be expressed specifically in passage cells. In order to understand the patterning, differentiation, and physiological role of passage it would be crucial to identify some genes that are exclusively expressed in these cells. In order to identify such genes, I first generated fluorescent marker lines of stele-expressed transporters that have been reported to be expressed in the passage cells. My aim was to first highlight the passage cells in a non-specific way. In order to find passage cell specific genes I then adapted a two-component system based on previously published methods for gene expression profiling of individual cell types. This approach will allow us to target only the passage cells and then to study gene expression specifically in this cell type. Taken together, this preparatory work will provide an entry point to understand the formation and role of endodermal passage cells. - Chez les plantes vasculaires, la caractéristique la plus commune des cellules différentiées de l'endoderme est la présence de cadres de Caspary. Cette structure correspond à une imprégnation localisée des parties transversales et anticlinales de la paroi cellulaire. Cela donne naissance, autour de la cellule, à un anneau/cadre qui est coordonné par rapport aux cellules voisines. De manière analogue aux jonctions serrées des épithéliums chez les animaux, les cadres de Caspary agissent chez les plantes comme barrière de diffusion, contrôlant le mouvement de l'eau et des ions à travers la racine entre le sol et la stèle. Depuis leur première description par Robert Caspary en 1865, beaucoup de tentatives ont eu pour but de définir la nature chimique de ces cadres de Caspary. Après l'étude de différentes espèces végétales, à la fois la subérine, la lignine ou les deux ont été revendiquées comme étant des composants importants de ces cadres. Malgré tout, leur nature chimique exacte est restée longtemps énigmatique. Cette controverse provient de la confusion et du manque de connaissance concernant la détermination précise des trois stades de développement de l'endoderme. Les cadres de Caspary représentent uniquement le stade primaire de différentiation de l'endoderme. Celui-ci est suivi par le second stade de différentiation, la déposition de lamelles de subérine tout autour de la cellule endodermal. De ce fait, l'analyse chimique de racines entières ou de cellules d'endoderme isolées ne permet pas de séparer les stades de différentiation primaire et secondaire et aboutit donc à la présence des deux polymères. Il est également crucial de clarifier ce point dans le but de connaître quelle machinerie cellulaire localisée à la paroi cellulaire permet l'élaboration des cadres de Caspary. En utilisant les connaissances et les outils accumulés récemment grâce à la plante modèle Arabidopsis thaliana, divers techniques histologiques et chimiques ont été utilisées dans cette étude. Un point particulier de mon travail a été de dégrader ou d'inhiber complètement la formation de lignine ou de subérine en utilisant des approches de génétique classique ou moléculaire. Le but étant d'observer l'effet de l'absence d'un de ces deux polymères sur la formation des cadres de Caspary et l'établissement d'une barrière de diffusion fonctionnelle. De manière frappante, le fait d'interférer avec la voie de biosynthèse de monolignol (monomères de lignine) abolit la formation des cadres de Caspary et retarde l'élaboration d'une barrière de diffusion fonctionnelle. Par contre, des plantes transgéniques dépourvues d'une quantité détectable de subérine sont quant à elles toujours capables de développer des cadres de Caspary fonctionnels. Mises en commun, ces expériences démontrent que le polymère formant les cadres de Caspary dans la partie jeune de la racine est fait de monolignol, et que de ce fait il s'agit de lignine. La subérine, quant à elle, est formée bien plus tard durant le développement de l'endoderme, de plus il s'agit d'une modification de la paroi secondaire. Ces cadres de Caspary précoces faits de lignine suffisent donc à bloquer la diffusion extracellulaire, contrairement à la subérine. De plus, la machinerie de biosynthèse de la subérine n'est pas encore présente au moment de la formation des cadres de Caspary. Notre étude permet donc de mettre un terme au long débat concernant la nature chimique des cadres de Caspary. De plus, elle ouvre la porte à de nouvelles approches dans la recherche sur la lignine, plus particulièrement pour identifier des composants permettant la déposition localisée de ce polymère dans la paroi cellulaire. J'ai aussi fais des efforts pour mettre en évidence la formation ainsi que le rôle des cellules de passage dans les jeunes racines. Dans la littérature, les cellules de passage sont définies comme de la cellule endodermal faisant face aux pôles xylèmes et dont la paroi n'est pas subérisée. Du fait que ces cellules contiennent uniquement des cadres de Caspary et pas de lamelle de subérine, il a été supposé qu'elles ne devraient offrir que peu de résistance au passage de l'eau et des nutriments entre le sol et la stèle. Le rôle de ces cellules de passage est toujours loin d'être clair, de plus aucun gène s'exprimant spécifiquement dans ces cellules n'a été découvert à ce jour. De manière à identifier de tels gènes, j'ai tout d'abord généré des marqueurs fluorescents pour des transporteurs exprimés dans la stèle mais dont l'expression avait également été signalée dans l'endoderme, uniquement dans les cellules de passage. J'ai ensuite développé un système à deux composants basé sur des méthodes déjà publiées, visant principalement à étudier le profil d'expression génique dans un type cellulaire donné. En recoupant les gènes exprimés spécifiquement dans l'endoderme à ceux exprimés dans la stèle et les cellules de passage, il nous sera possible d'identifier le transriptome spécifique de ces cellules. Pris dans leur ensemble, ces résultats devraient donner un bon point d'entrée dans la définition et la compréhension des cellules de passage.

Multiple refugia and barriers explain the phylogeography of the Valais shrew, Sorex antinorii (Mammalia: Soricomorpha), Sorex antinorii (Mammalia: Soricomorpha)

The aim of the present study was to investigate the genetic structure of the Valais shrew (Sorex antinorii) by a combined phylogeographical and landscape genetic approach, and thereby to infer the locations of glacial refugia and establish the influence of geographical barriers. We sequenced part of the mitochondrial cytochrome b (cyt b) gene of 179 individuals of S. antinorii sampled across the entire species' range. Six specimens attributed to S. arunchi were included in the analysis. The phylogeographical pattern was assessed by Bayesian molecular phylogenetic reconstruction, population genetic analyses, and a species distribution modelling (SDM)-based hindcasting approach. We also used landscape genetics (including isolation-by-resistance) to infer the determinants of current intra-specific genetic structure. The phylogeographical analysis revealed shallow divergence among haplotypes and no clear substructure within S. antinorii. The starlike structure of the median-joining network is consistent with population expansion from a single refugium, probably located in the Apennines. Long branches observed on the same network also suggest that another refugium may have existed in the north-eastern part of Italy. This result is consistent with SDM, which also suggests several habitable areas for S. antinorii in the Italian peninsula during the LGM. Therefore S. antinorii appears to have occupied disconnected glacial refugia in the Italian peninsula, supporting previous data for other species showing multiple refugia within southern refugial areas. By coupling genetic analyses and SDM, we were able to infer how past climatic suitability contributed to genetic divergence of populations. The genetic differentiation shown in the present study does not support the specific status of S. arunchi.

"I'll look it up on the Web first": barriers and overcoming barriers to consult for sexual dysfunction among young men.

QUESTIONS UNDER STUDY: Our aim was to identify the barriers young men face to consult a health professional when they encounter sexual dysfunctions and where they turn to, if so, for answers. METHODS: We conducted an exploratory qualitative research including 12 young men aged 16-20 years old seen in two focus groups. Discussions were triggered through vignettes about sexual dysfunction. RESULTS: Young men preferred not to talk about sexual dysfunction problems with anyone and to solve them alone as it is considered an intimate and embarrassing subject which can negatively impact their masculinity. Confidentiality appeared to be the most important criterion in disclosing an intimate subject to a health professional. Participants raised the problem of males' accessibility to services and lack of reason to consult. Two criteria to address the problem were if it was long-lasting or considered as physical. The Internet was unanimously considered as an initial solution to solve a problem, which could guide them to a face-to-face consultation if necessary. CONCLUSIONS: Results suggest that Internet-based tools should be developed to become an easy access door to sexual health services for young men. Wherever they consult and for whatever problem, sexual health must be on the agenda.

How the mind constitutes itself through perceptual learning

Most theories of perception assume a rigid relationship between objects of the physical world and the corresponding mental representations. We show by a priori reasoning that this assumption is not fulfilled. We claim instead that all object-representation correspondences have to be learned. However, we cannot learn to perceive all objects that there are in the world. We arrive at these conclusions by a combinatory analysis of a fictive stimulus world and the way to cope with its complexity, which is perceptual learning. We show that successful perceptual learning requires changes in the representational states of the brain that are not derived directly from the constitution of the physical world. The mind constitutes itself through perceptual learning.