Machine learning for geospatial data : algorithms, software tools and case studies

Résumé Cette thèse est consacrée à l'analyse, la modélisation et la visualisation de données environnementales à référence spatiale à l'aide d'algorithmes d'apprentissage automatique (Machine Learning). L'apprentissage automatique peut être considéré au sens large comme une sous-catégorie de l'intelligence artificielle qui concerne particulièrement le développement de techniques et d'algorithmes permettant à une machine d'apprendre à partir de données. Dans cette thèse, les algorithmes d'apprentissage automatique sont adaptés pour être appliqués à des données environnementales et à la prédiction spatiale. Pourquoi l'apprentissage automatique ? Parce que la majorité des algorithmes d'apprentissage automatiques sont universels, adaptatifs, non-linéaires, robustes et efficaces pour la modélisation. Ils peuvent résoudre des problèmes de classification, de régression et de modélisation de densité de probabilités dans des espaces à haute dimension, composés de variables informatives spatialisées (« géo-features ») en plus des coordonnées géographiques. De plus, ils sont idéaux pour être implémentés en tant qu'outils d'aide à la décision pour des questions environnementales allant de la reconnaissance de pattern à la modélisation et la prédiction en passant par la cartographie automatique. Leur efficacité est comparable au modèles géostatistiques dans l'espace des coordonnées géographiques, mais ils sont indispensables pour des données à hautes dimensions incluant des géo-features. Les algorithmes d'apprentissage automatique les plus importants et les plus populaires sont présentés théoriquement et implémentés sous forme de logiciels pour les sciences environnementales. Les principaux algorithmes décrits sont le Perceptron multicouches (MultiLayer Perceptron, MLP) - l'algorithme le plus connu dans l'intelligence artificielle, le réseau de neurones de régression généralisée (General Regression Neural Networks, GRNN), le réseau de neurones probabiliste (Probabilistic Neural Networks, PNN), les cartes auto-organisées (SelfOrganized Maps, SOM), les modèles à mixture Gaussiennes (Gaussian Mixture Models, GMM), les réseaux à fonctions de base radiales (Radial Basis Functions Networks, RBF) et les réseaux à mixture de densité (Mixture Density Networks, MDN). Cette gamme d'algorithmes permet de couvrir des tâches variées telle que la classification, la régression ou l'estimation de densité de probabilité. L'analyse exploratoire des données (Exploratory Data Analysis, EDA) est le premier pas de toute analyse de données. Dans cette thèse les concepts d'analyse exploratoire de données spatiales (Exploratory Spatial Data Analysis, ESDA) sont traités selon l'approche traditionnelle de la géostatistique avec la variographie expérimentale et selon les principes de l'apprentissage automatique. La variographie expérimentale, qui étudie les relations entre pairs de points, est un outil de base pour l'analyse géostatistique de corrélations spatiales anisotropiques qui permet de détecter la présence de patterns spatiaux descriptible par une statistique. L'approche de l'apprentissage automatique pour l'ESDA est présentée à travers l'application de la méthode des k plus proches voisins qui est très simple et possède d'excellentes qualités d'interprétation et de visualisation. Une part importante de la thèse traite de sujets d'actualité comme la cartographie automatique de données spatiales. Le réseau de neurones de régression généralisée est proposé pour résoudre cette tâche efficacement. Les performances du GRNN sont démontrées par des données de Comparaison d'Interpolation Spatiale (SIC) de 2004 pour lesquelles le GRNN bat significativement toutes les autres méthodes, particulièrement lors de situations d'urgence. La thèse est composée de quatre chapitres : théorie, applications, outils logiciels et des exemples guidés. Une partie importante du travail consiste en une collection de logiciels : Machine Learning Office. Cette collection de logiciels a été développée durant les 15 dernières années et a été utilisée pour l'enseignement de nombreux cours, dont des workshops internationaux en Chine, France, Italie, Irlande et Suisse ainsi que dans des projets de recherche fondamentaux et appliqués. Les cas d'études considérés couvrent un vaste spectre de problèmes géoenvironnementaux réels à basse et haute dimensionnalité, tels que la pollution de l'air, du sol et de l'eau par des produits radioactifs et des métaux lourds, la classification de types de sols et d'unités hydrogéologiques, la cartographie des incertitudes pour l'aide à la décision et l'estimation de risques naturels (glissements de terrain, avalanches). Des outils complémentaires pour l'analyse exploratoire des données et la visualisation ont également été développés en prenant soin de créer une interface conviviale et facile à l'utilisation. Machine Learning for geospatial data: algorithms, software tools and case studies Abstract The thesis is devoted to the analysis, modeling and visualisation of spatial environmental data using machine learning algorithms. In a broad sense machine learning can be considered as a subfield of artificial intelligence. It mainly concerns with the development of techniques and algorithms that allow computers to learn from data. In this thesis machine learning algorithms are adapted to learn from spatial environmental data and to make spatial predictions. Why machine learning? In few words most of machine learning algorithms are universal, adaptive, nonlinear, robust and efficient modeling tools. They can find solutions for the classification, regression, and probability density modeling problems in high-dimensional geo-feature spaces, composed of geographical space and additional relevant spatially referenced features. They are well-suited to be implemented as predictive engines in decision support systems, for the purposes of environmental data mining including pattern recognition, modeling and predictions as well as automatic data mapping. They have competitive efficiency to the geostatistical models in low dimensional geographical spaces but are indispensable in high-dimensional geo-feature spaces. The most important and popular machine learning algorithms and models interesting for geo- and environmental sciences are presented in details: from theoretical description of the concepts to the software implementation. The main algorithms and models considered are the following: multi-layer perceptron (a workhorse of machine learning), general regression neural networks, probabilistic neural networks, self-organising (Kohonen) maps, Gaussian mixture models, radial basis functions networks, mixture density networks. This set of models covers machine learning tasks such as classification, regression, and density estimation. Exploratory data analysis (EDA) is initial and very important part of data analysis. In this thesis the concepts of exploratory spatial data analysis (ESDA) is considered using both traditional geostatistical approach such as_experimental variography and machine learning. Experimental variography is a basic tool for geostatistical analysis of anisotropic spatial correlations which helps to understand the presence of spatial patterns, at least described by two-point statistics. A machine learning approach for ESDA is presented by applying the k-nearest neighbors (k-NN) method which is simple and has very good interpretation and visualization properties. Important part of the thesis deals with a hot topic of nowadays, namely, an automatic mapping of geospatial data. General regression neural networks (GRNN) is proposed as efficient model to solve this task. Performance of the GRNN model is demonstrated on Spatial Interpolation Comparison (SIC) 2004 data where GRNN model significantly outperformed all other approaches, especially in case of emergency conditions. The thesis consists of four chapters and has the following structure: theory, applications, software tools, and how-to-do-it examples. An important part of the work is a collection of software tools - Machine Learning Office. Machine Learning Office tools were developed during last 15 years and was used both for many teaching courses, including international workshops in China, France, Italy, Ireland, Switzerland and for realizing fundamental and applied research projects. Case studies considered cover wide spectrum of the real-life low and high-dimensional geo- and environmental problems, such as air, soil and water pollution by radionuclides and heavy metals, soil types and hydro-geological units classification, decision-oriented mapping with uncertainties, natural hazards (landslides, avalanches) assessments and susceptibility mapping. Complementary tools useful for the exploratory data analysis and visualisation were developed as well. The software is user friendly and easy to use.

Patterns of protein and carbohydrate accumulation during somatic embryogenesis of Acca sellowiana

The aim of this work was to quantify the protein, starch and total sugars levels during histodifferentiation and development of somatic embryos of Acca sellowiana Berg. For histological observations, the samples were dehydrated in a battery of ethanol, embedded in historesin and stained with toluidine blue (morphology), coomassie blue (protein bodies) and periodic acid-Schiff (starch). Proteins were extracted using a buffer solution, precipitated using ethanol and quantified using the Bradford reagent. Total sugars were extracted using a methanol-chloroform-water (12:5:3) solution and quantified by a reaction with anthrone at 0.2%. Starch was extracted using a 30% perchloric acid solution and quantified by a reaction with anthrone at 0.2%. During the somatic embryogenesis' in vitro morphogenesis and differentiation processes, the total protein levels decreased and the soluble sugars levels increased during the first 30 days in culture and remained stable until the 120th day. On the other hand, total protein levels increased according to the progression in the developmental stages of the somatic embryos. The levels of total sugars and starch increased in the heart and cotyledonary stages, and decreased in the torpedo and pre-cotyledonary stages. These compounds play a central role in the development of somatic embryos of Acca sellowiana.

Impact of the Advisa MRI pacing system on the diagnostic quality of cardiac MR images and contraction patterns of cardiac muscle during scans: Advisa MRI randomized clinical multicenter study results.

BACKGROUND: The Advisa MRI system is designed to safely undergo magnetic resonance imaging (MRI). Its influence on image quality is not well known. OBJECTIVE: To evaluate cardiac magnetic resonance (CMR) image quality and to characterize myocardial contraction patterns by using the Advisa MRI system. METHODS: In this international trial with 35 participating centers, an Advisa MRI system was implanted in 263 patients. Of those, 177 were randomized to the MRI group and 150 underwent MRI scans at the 9-12-week visit. Left ventricular (LV) and right ventricular (RV) cine long-axis steady-state free precession MR images were graded for quality. Signal loss along the implantable pulse generator and leads was measured. The tagging CMR data quality was assessed as the percentage of trackable tagging points on complementary spatial modulation of magnetization acquisitions (n=16) and segmental circumferential fiber shortening was quantified. RESULTS: Of all cine long-axis steady-state free precession acquisitions, 95% of LV and 98% of RV acquisitions were of diagnostic quality, with 84% and 93%, respectively, being of good or excellent quality. Tagging points were trackable from systole into early diastole (360-648 ms after the R-wave) in all segments. During RV pacing, tagging demonstrated a dyssynchronous contraction pattern, which was not observed in nonpaced (n = 4) and right atrial-paced (n = 8) patients. CONCLUSIONS: In the Advisa MRI study, high-quality CMR images for the assessment of cardiac anatomy and function were obtained in most patients with an implantable pacing system. In addition, this study demonstrated the feasibility of acquiring tagging data to study the LV function during pacing.

Estimating water application efficiency for drip irrigation emitter patterns on banana

The objective of this work was to evaluate root and water distribution in irrigated banana (Musa sp.), in order to determine the water application efficiency for different drip irrigation emitter patterns. Three drip emitter patterns were studied: two 4-L h-1 emitters per plant (T1), four 4-L h-1 emitters per plant (T2), and five 4-L h-1 emitters per plant (T3). The emitters were placed in a lateral line. In the treatment T3, the emitters formed a continuous strip. The cultivated area used was planted with banana cultivar BRS Tropical, with a 3-m spacing between rows and a 2.5-m spacing between plants. Soil moisture and root length data were collected during the first production cycle at five radial distances and depths, in a 0.20x0.20 m vertical grid. The experiment was carried out in a sandy clay loam Xanthic Hapludox. Soil moisture data were collected every 10 min for a period of five days using TDR probes. Water application efficiency was of 83, 88 and 92% for the systems with two, four and five emitters per plant, respectively. It was verified that an increase in the number of emitters in the lateral line promoted better root distribution, higher water extraction, and less deep percolation losses.

Distribution and agroclimatic characterization of potential cultivation regions of physic nut in Mexico

The purposes of this study were to determine the distribution and climatic patterns of current and future physic nut (Jatropha curcas) cultivation regions in Mexico, and to identify possible locations for in vivo germplasm banks establishment, using geographic information systems. Current climatic data were processed by Floramap software to obtain distribution maps and climatic patterns of regions where wild physic nuts could be found. DIVA-GIS software analyzed current climatic data (Worldclim model) and climatic data generated by CCM3 model to identify current and future physic nut cultivation regions, respectively. The distribution map showed that physic nut was present in most of the tropical and subtropical areas of Mexico, which corresponded to three agroclimatic regions. Climate types were Aw2, Aw1, and Bs1, for regions 1, 2 and 3, respectively. Nontoxic genotypes were associated with region 2, and toxic genotypes were associated with regions 1 and 3. According to the current and future cultivation regions identified, the best suitable ones to establish in vivo germplasm collections were the coast of Michoacán and the Isthmus of Tehuantepec, located among the states of Veracruz, Oaxaca and Chiapas.

Do geographic distribution, niche property and life form explain plants' vulnerability to global change?

We modelled the future distribution in 2050 of 975 endemic plant species in southern Africa distributed among seven life forms, including new methodological insights improving the accuracy and ecological realism of predictions of global changes studies by: (i) using only endemic species as a way to capture the full realized niche of species, (ii) considering the direct impact of human pressure on landscape and biodiversity jointly with climate, and (iii) taking species' migration into account. Our analysis shows important promises for predicting the impacts of climate change in conjunction with land transformation. We have shown that the endemic flora of Southern Africa on average decreases with 41% in species richness among habitats and with 39% on species distribution range for the most optimistic scenario. We also compared the patterns of species' sensitivity with global change across life forms, using ecological and geographic characteristics of species. We demonstrate here that species and life form vulnerability to global changes can be partly explained according to species' (i) geographical distribution along climatic and biogeographic gradients, like climate anomalies, (ii) niche breadth or (iii) proximity to barrier preventing migration. Our results confirm that the sensitivity of a given species to global environmental changes depends upon its geographical distribution and ecological proprieties, and makes it possible to estimate a priori its potential sensitivity to these changes.

Phytogeographical origin of Madeiran common beans based on phaseolin patterns

The objective of this work was to determine the geographic origin of the Madeiran common bean (Phaseolus vulgaris) gene pool. Phaseolin patterns of 50 accessions representing the diversity of common bean collected in Madeira, Portugal, and conserved in the ISOPlexis Germplasm Bank, were analysed using the Experion automated electrophoresis system, based on lab-on-a-chip technology. Five common bean standard varieties with typical phaseolin patterns were used to determine the phytogeographical origin of the Madeiran common bean accessions. Ninety two percent of the accessions exhibited a phaseolin pattern consistent with the one of common bean types belonging to the Andean gene pool, while the origin of the remaining 8% of the accessions was indistinguishable. The application of a similarity coefficient of 85%, based on Pearson correlations, increases the number of accessions with uncertain pattern. The analytical approach used permitted the determination of the origin of the common bean gene pool, which is Andean in 98% of the cases, and clustering of the observed variability among the Madeiran common beans.

Face Short-Term Memory-Related Electroencephalographic Patterns can Differentiate Multi- versus Single-Domain Amnestic Mild Cognitive Impairment.

Amnestic mild cognitive impairment (aMCI) is characterized by memory deficits alone (single-domain, sd-aMCI) or associated with other cognitive disabilities (multi-domain, md-aMCI). The present study assessed the patterns of electroencephalographic (EEG) activity during the encoding and retrieval phases of short-term memory in these two aMCI subtypes, to identify potential functional differences according to the neuropsychological profile. Continuous EEG was recorded in 43 aMCI patients, whose 16 sd-aMCI and 27 md-aMCI, and 36 age-matched controls (EC) during delayed match-to-sample tasks for face and letter stimuli. At encoding, attended stimuli elicited parietal alpha (8-12 Hz) power decrease (desynchronization), whereas distracting stimuli were associated with alpha power increase (synchronization) over right central sites. No difference was observed in parietal alpha desynchronization among the three groups. For attended faces, the alpha synchronization underlying suppression of distracting letters was reduced in both aMCI subgroups, but more severely in md-aMCI cases that differed significantly from EC. At retrieval, the early N250r recognition effect was significantly reduced for faces in md-aMCI as compared to both sd-aMCI and EC. The results suggest a differential alteration of working memory cerebral processes for faces in the two aMCI subtypes, face covert recognition processes being specifically altered in md-aMCI.

Clypeal patterning in the paper wasp Polistes dominulus: no evidence of adaptive value in the wild

Status signals function in a number of species to communicate competitive ability to conspecific rivals during competition for resources. In the paper wasp Polistes dominulus, variable black clypeal patterns are thought to be important in mediating competition among females. Results of previous behavioral experiments in the lab indicate that P dominulus clypeal patterns provide information about an individual's competitive ability to rivals during agonistic interactions. To date, however, there has been no detailed examination of the adaptive value of clypeal patterns in the wild. To address this, we looked for correlations between clypeal patterning and various fitness measures, including reproductive success, hierarchical rank, and survival, in a large, free-living population of P. dominulus in southern Spain. Reproductive success over the nesting season was not correlated with clypeal patterning. Furthermore, there was no relationship between a female's clypeal patterning and the rank she achieved within the hierarchy or her survival during nest founding. Overall, we found no evidence that P dominulus clypeal patterns are related to competitive ability or other aspects of quality in our population. This result is consistent with geographical variation in the adaptive value of clypeal patterns between P. dominulus populations; however, data on the relationship between patterning and fitness from other populations are required to test this hypothesis.

Genetic diversity for gliadin patterns of durum wheat landraces in the Northwest of Iran and Azerbaijan

The objective of this study was to identify gliadin band patterns and the extent of genetic diversity in durum wheat genotypes from Northwestern Iran and the Republic of Azerbaijan. Gliadins from 46 landraces and four cultivars were evaluated through acid PAGE analyses. Sixty-six polymorphic bands and 81 patterns were identified. Twenty-four different motility bands and 22 patterns were found in the ω gliadin region with 14 polymorph bands and 20 patterns for α and γ gliadins, and 14 bands and 19 different patterns for β gliadins. The combination of these patterns generated 38 and 39 combinations for Gli-1 and Gli-2 loci, respectively. The genetic diversity index (H) was higher for α gliadins (0.924), followed by ω and γ gliadins (0.899 and 0.878, respectively), and for β gliadin patterns (0.866). Extensive polymorphism (H = 0.875) was observed in four gliadin pattern regions, with higher genetic diversity in the Iranian landraces than in the Azerbaijani ones. Each genotype had special identifying patterns in the gliadin acid PAGE analysis, and cluster analysis based on Jaccard's similarity coefficients formed six groups. Gliadin has a simple, repeatable and economic analysis, and can be used in genetic studies

Comment on 'The Molecular Evolutionary Patterns of the Insulin/FOXO Signaling Pathway'

Letter to the Editor on Wang M, Wang Q, Wang Z, Zhang X, Pan Y. The molecular evolutionary patterns of the insulin/FOXO signaling pathway

Recent patterns in gastric cancer: a global overview.

Until the mid-1990s, gastric cancer has been the first cause of cancer death worldwide, although rates had been declining for several decades and gastric cancer has become a relatively rare cancer in North America and in most Northern and Western Europe, but not in Eastern Europe, Russia and selected areas of Central and South America or East Asia. We analyzed gastric cancer mortality in Europe and other areas of the world from 1980 to 2005 using joinpoint regression analysis, and provided updated site-specific incidence rates from 51 selected registries. Over the last decade, the annual percent change (APC) in mortality rate was around -3, -4% for the major European countries. The APC were similar for the Republic of Korea (APC = -4.3%), Australia (-3.7%), the USA (-3.6%), Japan (-3.5%), Ukraine (-3%) and the Russian Federation (-2.8%). In Latin America, the decline was less marked, but constant with APC around -1.6% in Chile and Brazil, -2.3% in Argentina and Mexico and -2.6% in Colombia. Cancers in the fundus and pylorus are more common in high incidence and mortality areas and have been declining more than cardia gastric cancer. Steady downward trends persist in gastric cancer mortality worldwide even in middle aged population, and hence further appreciable declines are likely in the near future.