Three essays on major trends in a slow clockspeed industry : the case of industrial automation

The motivation for this research initiated from the abrupt rise and fall of minicomputers which were initially used both for industrial automation and business applications due to their significantly lower cost than their predecessors, the mainframes. Later industrial automation developed its own vertically integrated hardware and software to address the application needs of uninterrupted operations, real-time control and resilience to harsh environmental conditions. This has led to the creation of an independent industry, namely industrial automation used in PLC, DCS, SCADA and robot control systems. This industry employs today over 200'000 people in a profitable slow clockspeed context in contrast to the two mainstream computing industries of information technology (IT) focused on business applications and telecommunications focused on communications networks and hand-held devices. Already in 1990s it was foreseen that IT and communication would merge into one Information and communication industry (ICT). The fundamental question of the thesis is: Could industrial automation leverage a common technology platform with the newly formed ICT industry? Computer systems dominated by complex instruction set computers (CISC) were challenged during 1990s with higher performance reduced instruction set computers (RISC). RISC started to evolve parallel to the constant advancement of Moore's law. These developments created the high performance and low energy consumption System-on-Chip architecture (SoC). Unlike to the CISC processors RISC processor architecture is a separate industry from the RISC chip manufacturing industry. It also has several hardware independent software platforms consisting of integrated operating system, development environment, user interface and application market which enables customers to have more choices due to hardware independent real time capable software applications. An architecture disruption merged and the smartphone and tablet market were formed with new rules and new key players in the ICT industry. Today there are more RISC computer systems running Linux (or other Unix variants) than any other computer system. The astonishing rise of SoC based technologies and related software platforms in smartphones created in unit terms the largest installed base ever seen in the history of computers and is now being further extended by tablets. An underlying additional element of this transition is the increasing role of open source technologies both in software and hardware. This has driven the microprocessor based personal computer industry with few dominating closed operating system platforms into a steep decline. A significant factor in this process has been the separation of processor architecture and processor chip production and operating systems and application development platforms merger into integrated software platforms with proprietary application markets. Furthermore the pay-by-click marketing has changed the way applications development is compensated: Three essays on major trends in a slow clockspeed industry: The case of industrial automation 2014 freeware, ad based or licensed - all at a lower price and used by a wider customer base than ever before. Moreover, the concept of software maintenance contract is very remote in the app world. However, as a slow clockspeed industry, industrial automation has remained intact during the disruptions based on SoC and related software platforms in the ICT industries. Industrial automation incumbents continue to supply systems based on vertically integrated systems consisting of proprietary software and proprietary mainly microprocessor based hardware. They enjoy admirable profitability levels on a very narrow customer base due to strong technology-enabled customer lock-in and customers' high risk leverage as their production is dependent on fault-free operation of the industrial automation systems. When will this balance of power be disrupted? The thesis suggests how industrial automation could join the mainstream ICT industry and create an information, communication and automation (ICAT) industry. Lately the Internet of Things (loT) and weightless networks, a new standard leveraging frequency channels earlier occupied by TV broadcasting, have gradually started to change the rigid world of Machine to Machine (M2M) interaction. It is foreseeable that enough momentum will be created that the industrial automation market will in due course face an architecture disruption empowered by these new trends. This thesis examines the current state of industrial automation subject to the competition between the incumbents firstly through a research on cost competitiveness efforts in captive outsourcing of engineering, research and development and secondly researching process re- engineering in the case of complex system global software support. Thirdly we investigate the industry actors', namely customers, incumbents and newcomers, views on the future direction of industrial automation and conclude with our assessments of the possible routes industrial automation could advance taking into account the looming rise of the Internet of Things (loT) and weightless networks. Industrial automation is an industry dominated by a handful of global players each of them focusing on maintaining their own proprietary solutions. The rise of de facto standards like IBM PC, Unix and Linux and SoC leveraged by IBM, Compaq, Dell, HP, ARM, Apple, Google, Samsung and others have created new markets of personal computers, smartphone and tablets and will eventually also impact industrial automation through game changing commoditization and related control point and business model changes. This trend will inevitably continue, but the transition to a commoditized industrial automation will not happen in the near future.

Intracellular nanomanipulation by a photonic-force microscope with real-time acquisition of a 3D stiffness matrix.

A traditional photonic-force microscope (PFM) results in huge sets of data, which requires tedious numerical analysis. In this paper, we propose instead an analog signal processor to attain real-time capabilities while retaining the richness of the traditional PFM data. Our system is devoted to intracellular measurements and is fully interactive through the use of a haptic joystick. Using our specialized analog hardware along with a dedicated algorithm, we can extract the full 3D stiffness matrix of the optical trap in real time, including the off-diagonal cross-terms. Our system is also capable of simultaneously recording data for subsequent offline analysis. This allows us to check that a good correlation exists between the classical analysis of stiffness and our real-time measurements. We monitor the PFM beads using an optical microscope. The force-feedback mechanism of the haptic joystick helps us in interactively guiding the bead inside living cells and collecting information from its (possibly anisotropic) environment. The instantaneous stiffness measurements are also displayed in real time on a graphical user interface. The whole system has been built and is operational; here we present early results that confirm the consistency of the real-time measurements with offline computations.

MyHits: improvements to an interactive resource for analyzing protein sequences.

The MyHits web site (http://myhits.isb-sib.ch) is an integrated service dedicated to the analysis of protein sequences. Since its first description in 2004, both the user interface and the back end of the server were improved. A number of tools (e.g. MAFFT, Jacop, Dotlet, Jalview, ESTScan) were added or updated to improve the usability of the service. The MySQL schema and its associated API were revamped and the database engine (HitKeeper) was separated from the web interface. This paper summarizes the current status of the server, with an emphasis on the new services.

A rapid and effective vignetting correction for quantitative microscopy

Images acquired using optical microscopes are inherently subject to vignetting effects due to imperfect illumination and image acquisition. However, such vignetting effects hamper accurate extraction of quantitative information from biological images, leading to less effective image segmentation and increased noise in the measurements. Here, we describe a rapid and effective method for vignetting correction, which generates an estimate for a correction function from the background fluorescence without the need to acquire additional calibration images. We validate the usefulness of this algorithm using artificially distorted images as a gold standard for assessing the accuracy of the applied correction and then demonstrate that this correction method enables the reliable detection of biologically relevant variation in cell populations. A simple user interface called FlattifY was developed and integrated into the image analysis platform YeastQuant to facilitate easy application of vignetting correction to a wide range of images.

Automated scoring of AFLPs using RawGeno v 2.0, a free R CRAN library.

Amplified Fragment Length Polymorphisms (AFLPs) are a cheap and efficient protocol for generating large sets of genetic markers. This technique has become increasingly used during the last decade in various fields of biology, including population genomics, phylogeography, and genome mapping. Here, we present RawGeno, an R library dedicated to the automated scoring of AFLPs (i.e., the coding of electropherogram signals into ready-to-use datasets). Our program includes a complete suite of tools for binning, editing, visualizing, and exporting results obtained from AFLP experiments. RawGeno can either be used with command lines and program analysis routines or through a user-friendly graphical user interface. We describe the whole RawGeno pipeline along with recommendations for (a) setting the analysis of electropherograms in combination with PeakScanner, a program freely distributed by Applied Biosystems; (b) performing quality checks; (c) defining bins and proceeding to scoring; (d) filtering nonoptimal bins; and (e) exporting results in different formats.

Bridge to life: the Lifebridge B2T extracorporeal life support system in an in vitro trial.

Extracorporeal life support systems (ECLS) have become common in cardiothoracic surgery, but are still "Terra Incognita" in other medical fields due to the fact that perfusion units are normally bound to cardiothoracic centres. The Lifebridge B2T is an ECLS that is meant to be used as an easy and fast-track extracorporeal cardiac support to provide short-term perfusion for the transport of a patient to a specialized centre. With the Lifebridge B2T it is now possible to provide extracorporeal bypass for patients in hospitals without a perfusion unit. The Lifebridge B2T was tested on three calves to analyze the handling, performance and security of this system. The Lifebridge B2T safely can be used clinically and can provide full extracorporeal support for patients in cardiac or pulmonary failure. Flows up to 3.9 +/- 0.2l/min were reached, with an inflow pressure of -103 +/- 13mmHg, using a 21Fr. BioMedicus (Medtronic, Minneapolis, MN, USA) venous cannula. The "Plug and Play" philosophy, with semi-automatic priming, integrated check-list, a long battery time of over two hours and instinctively designed user interface, makes this device very interesting for units with high-risk interventions, such as catheterisation labs. If a system is necessary in an emergency unit, the Lifebridge can provide a high security level, even in centres not acquainted with cardiopulmonary bypass.

ConnectomeViewer - Multi-Modal Multi-Level Network Visualization and Analysis.

Introduction: The field of Connectomic research is growing rapidly, resulting from methodological advances in structural neuroimaging on many spatial scales. Especially progress in Diffusion MRI data acquisition and processing made available macroscopic structural connectivity maps in vivo through Connectome Mapping Pipelines (Hagmann et al, 2008) into so-called Connectomes (Hagmann 2005, Sporns et al, 2005). They exhibit both spatial and topological information that constrain functional imaging studies and are relevant in their interpretation. The need for a special-purpose software tool for both clinical researchers and neuroscientists to support investigations of such connectome data has grown. Methods: We developed the ConnectomeViewer, a powerful, extensible software tool for visualization and analysis in connectomic research. It uses the novel defined container-like Connectome File Format, specifying networks (GraphML), surfaces (Gifti), volumes (Nifti), track data (TrackVis) and metadata. Usage of Python as programming language allows it to by cross-platform and have access to a multitude of scientific libraries. Results: Using a flexible plugin architecture, it is possible to enhance functionality for specific purposes easily. Following features are already implemented: * Ready usage of libraries, e.g. for complex network analysis (NetworkX) and data plotting (Matplotlib). More brain connectivity measures will be implemented in a future release (Rubinov et al, 2009). * 3D View of networks with node positioning based on corresponding ROI surface patch. Other layouts possible. * Picking functionality to select nodes, select edges, get more node information (ConnectomeWiki), toggle surface representations * Interactive thresholding and modality selection of edge properties using filters * Arbitrary metadata can be stored for networks, thereby allowing e.g. group-based analysis or meta-analysis. * Python Shell for scripting. Application data is exposed and can be modified or used for further post-processing. * Visualization pipelines using filters and modules can be composed with Mayavi (Ramachandran et al, 2008). * Interface to TrackVis to visualize track data. Selected nodes are converted to ROIs for fiber filtering The Connectome Mapping Pipeline (Hagmann et al, 2008) processed 20 healthy subjects into an average Connectome dataset. The Figures show the ConnectomeViewer user interface using this dataset. Connections are shown that occur in all 20 subjects. The dataset is freely available from the homepage (connectomeviewer.org). Conclusions: The ConnectomeViewer is a cross-platform, open-source software tool that provides extensive visualization and analysis capabilities for connectomic research. It has a modular architecture, integrates relevant datatypes and is completely scriptable. Visit www.connectomics.org to get involved as user or developer.

CR1Dmod: A Matlab program to model 1D complex resistivity effects in electrical and electromagnetic surveys

We have constructed a forward modelling code in Matlab, capable of handling several commonly used electrical and electromagnetic methods in a 1D environment. We review the implemented electromagnetic field equations for grounded wires, frequency and transient soundings and present new solutions in the case of a non-magnetic first layer. The CR1Dmod code evaluates the Hankel transforms occurring in the field equations using either the Fast Hankel Transform based on digital filter theory, or a numerical integration scheme applied between the zeros of the Bessel function. A graphical user interface allows easy construction of 1D models and control of the parameters. Modelling results are in agreement with other authors, but the time of computation is less efficient than other available codes. Nevertheless, the CR1Dmod routine handles complex resistivities and offers solutions based on the full EM-equations as well as the quasi-static approximation. Thus, modelling of effects based on changes in the magnetic permeability and the permittivity is also possible.

Linking landscape- and population ecology for large population management modelling: the case of Ibex (capra ibex) in Switzerland

La biologie de la conservation est communément associée à la protection de petites populations menacées d?extinction. Pourtant, il peut également être nécessaire de soumettre à gestion des populations surabondantes ou susceptibles d?une trop grande expansion, dans le but de prévenir les effets néfastes de la surpopulation. Du fait des différences tant quantitatives que qualitatives entre protection des petites populations et contrôle des grandes, il est nécessaire de disposer de modèles et de méthodes distinctes. L?objectif de ce travail a été de développer des modèles prédictifs de la dynamique des grandes populations, ainsi que des logiciels permettant de calculer les paramètres de ces modèles et de tester des scénarios de gestion. Le cas du Bouquetin des Alpes (Capra ibex ibex) - en forte expansion en Suisse depuis sa réintroduction au début du XXème siècle - servit d?exemple. Cette tâche fut accomplie en trois étapes : En premier lieu, un modèle de dynamique locale, spécifique au Bouquetin, fut développé : le modèle sous-jacent - structuré en classes d?âge et de sexe - est basé sur une matrice de Leslie à laquelle ont été ajoutées la densité-dépendance, la stochasticité environnementale et la chasse de régulation. Ce modèle fut implémenté dans un logiciel d?aide à la gestion - nommé SIM-Ibex - permettant la maintenance de données de recensements, l?estimation automatisée des paramètres, ainsi que l?ajustement et la simulation de stratégies de régulation. Mais la dynamique d?une population est influencée non seulement par des facteurs démographiques, mais aussi par la dispersion et la colonisation de nouveaux espaces. Il est donc nécessaire de pouvoir modéliser tant la qualité de l?habitat que les obstacles à la dispersion. Une collection de logiciels - nommée Biomapper - fut donc développée. Son module central est basé sur l?Analyse Factorielle de la Niche Ecologique (ENFA) dont le principe est de calculer des facteurs de marginalité et de spécialisation de la niche écologique à partir de prédicteurs environnementaux et de données d?observation de l?espèce. Tous les modules de Biomapper sont liés aux Systèmes d?Information Géographiques (SIG) ; ils couvrent toutes les opérations d?importation des données, préparation des prédicteurs, ENFA et calcul de la carte de qualité d?habitat, validation et traitement des résultats ; un module permet également de cartographier les barrières et les corridors de dispersion. Le domaine d?application de l?ENFA fut exploré par le biais d?une distribution d?espèce virtuelle. La comparaison à une méthode couramment utilisée pour construire des cartes de qualité d?habitat, le Modèle Linéaire Généralisé (GLM), montra qu?elle était particulièrement adaptée pour les espèces cryptiques ou en cours d?expansion. Les informations sur la démographie et le paysage furent finalement fusionnées en un modèle global. Une approche basée sur un automate cellulaire fut choisie, tant pour satisfaire aux contraintes du réalisme de la modélisation du paysage qu?à celles imposées par les grandes populations : la zone d?étude est modélisée par un pavage de cellules hexagonales, chacune caractérisée par des propriétés - une capacité de soutien et six taux d?imperméabilité quantifiant les échanges entre cellules adjacentes - et une variable, la densité de la population. Cette dernière varie en fonction de la reproduction et de la survie locale, ainsi que de la dispersion, sous l?influence de la densité-dépendance et de la stochasticité. Un logiciel - nommé HexaSpace - fut développé pour accomplir deux fonctions : 1° Calibrer l?automate sur la base de modèles de dynamique (par ex. calculés par SIM-Ibex) et d?une carte de qualité d?habitat (par ex. calculée par Biomapper). 2° Faire tourner des simulations. Il permet d?étudier l?expansion d?une espèce envahisseuse dans un paysage complexe composé de zones de qualité diverses et comportant des obstacles à la dispersion. Ce modèle fut appliqué à l?histoire de la réintroduction du Bouquetin dans les Alpes bernoises (Suisse). SIM-Ibex est actuellement utilisé par les gestionnaires de la faune et par les inspecteurs du gouvernement pour préparer et contrôler les plans de tir. Biomapper a été appliqué à plusieurs espèces (tant végétales qu?animales) à travers le Monde. De même, même si HexaSpace fut initialement conçu pour des espèces animales terrestres, il pourrait aisément être étndu à la propagation de plantes ou à la dispersion d?animaux volants. Ces logiciels étant conçus pour, à partir de données brutes, construire un modèle réaliste complexe, et du fait qu?ils sont dotés d?une interface d?utilisation intuitive, ils sont susceptibles de nombreuses applications en biologie de la conservation. En outre, ces approches peuvent également s?appliquer à des questions théoriques dans les domaines de l?écologie des populations et du paysage.<br/><br/>Conservation biology is commonly associated to small and endangered population protection. Nevertheless, large or potentially large populations may also need human management to prevent negative effects of overpopulation. As there are both qualitative and quantitative differences between small population protection and large population controlling, distinct methods and models are needed. The aim of this work was to develop theoretical models to predict large population dynamics, as well as computer tools to assess the parameters of these models and to test management scenarios. The alpine Ibex (Capra ibex ibex) - which experienced a spectacular increase since its reintroduction in Switzerland at the beginning of the 20th century - was used as paradigm species. This task was achieved in three steps: A local population dynamics model was first developed specifically for Ibex: the underlying age- and sex-structured model is based on a Leslie matrix approach with addition of density-dependence, environmental stochasticity and culling. This model was implemented into a management-support software - named SIM-Ibex - allowing census data maintenance, parameter automated assessment and culling strategies tuning and simulating. However population dynamics is driven not only by demographic factors, but also by dispersal and colonisation of new areas. Habitat suitability and obstacles modelling had therefore to be addressed. Thus, a software package - named Biomapper - was developed. Its central module is based on the Ecological Niche Factor Analysis (ENFA) whose principle is to compute niche marginality and specialisation factors from a set of environmental predictors and species presence data. All Biomapper modules are linked to Geographic Information Systems (GIS); they cover all operations of data importation, predictor preparation, ENFA and habitat suitability map computation, results validation and further processing; a module also allows mapping of dispersal barriers and corridors. ENFA application domain was then explored by means of a simulated species distribution. It was compared to a common habitat suitability assessing method, the Generalised Linear Model (GLM), and was proven better suited for spreading or cryptic species. Demography and landscape informations were finally merged into a global model. To cope with landscape realism and technical constraints of large population modelling, a cellular automaton approach was chosen: the study area is modelled by a lattice of hexagonal cells, each one characterised by a few fixed properties - a carrying capacity and six impermeability rates quantifying exchanges between adjacent cells - and one variable, population density. The later varies according to local reproduction/survival and dispersal dynamics, modified by density-dependence and stochasticity. A software - named HexaSpace - was developed, which achieves two functions: 1° Calibrating the automaton on the base of local population dynamics models (e.g., computed by SIM-Ibex) and a habitat suitability map (e.g. computed by Biomapper). 2° Running simulations. It allows studying the spreading of an invading species across a complex landscape made of variously suitable areas and dispersal barriers. This model was applied to the history of Ibex reintroduction in Bernese Alps (Switzerland). SIM-Ibex is now used by governmental wildlife managers to prepare and verify culling plans. Biomapper has been applied to several species (both plants and animals) all around the World. In the same way, whilst HexaSpace was originally designed for terrestrial animal species, it could be easily extended to model plant propagation or flying animals dispersal. As these softwares were designed to proceed from low-level data to build a complex realistic model and as they benefit from an intuitive user-interface, they may have many conservation applications. Moreover, theoretical questions in the fields of population and landscape ecology might also be addressed by these approaches.

Supporting Business Model Modelling: a Compromise between Creativity and Constraints

Diagrams and tools help to support task modelling in engi- neering and process management. Unfortunately they are unfit to help in a business context at a strategic level, because of the flexibility needed for creative thinking and user friendly interactions. We propose a tool which bridges the gap between freedom of actions, encouraging creativity, and constraints, allowing validation and advanced features.

EuroDia: a beta-cell gene expression resource.

Type 2 diabetes mellitus (T2DM) is a major disease affecting nearly 280 million people worldwide. Whilst the pathophysiological mechanisms leading to disease are poorly understood, dysfunction of the insulin-producing pancreatic beta-cells is key event for disease development. Monitoring the gene expression profiles of pancreatic beta-cells under several genetic or chemical perturbations has shed light on genes and pathways involved in T2DM. The EuroDia database has been established to build a unique collection of gene expression measurements performed on beta-cells of three organisms, namely human, mouse and rat. The Gene Expression Data Analysis Interface (GEDAI) has been developed to support this database. The quality of each dataset is assessed by a series of quality control procedures to detect putative hybridization outliers. The system integrates a web interface to several standard analysis functions from R/Bioconductor to identify differentially expressed genes and pathways. It also allows the combination of multiple experiments performed on different array platforms of the same technology. The design of this system enables each user to rapidly design a custom analysis pipeline and thus produce their own list of genes and pathways. Raw and normalized data can be downloaded for each experiment. The flexible engine of this database (GEDAI) is currently used to handle gene expression data from several laboratory-run projects dealing with different organisms and platforms. Database URL: http://eurodia.vital-it.ch.

Development and experimental validation of a finite element model of total ankle replacement.

Total ankle replacement remains a less satisfactory solution compared to other joint replacements. The goal of this study was to develop and validate a finite element model of total ankle replacement, for future testing of hypotheses related to clinical issues. To validate the finite element model, an experimental setup was specifically developed and applied on 8 cadaveric tibias. A non-cemented press fit tibial component of a mobile bearing prosthesis was inserted into the tibias. Two extreme anterior and posterior positions of the mobile bearing insert were considered, as well as a centered one. An axial force of 2kN was applied for each insert position. Strains were measured on the bone surface using digital image correlation. Tibias were CT scanned before implantation, after implantation, and after mechanical tests and removal of the prosthesis. The finite element model replicated the experimental setup. The first CT was used to build the geometry and evaluate the mechanical properties of the tibias. The second CT was used to set the implant position. The third CT was used to assess the bone-implant interface conditions. The coefficient of determination (R-squared) between the measured and predicted strains was 0.91. Predicted bone strains were maximal around the implant keel, especially at the anterior and posterior ends. The finite element model presented here is validated for future tests using more physiological loading conditions.

Broadening the horizon--level 2.5 of the HUPO-PSI format for molecular interactions.

BACKGROUND: Molecular interaction Information is a key resource in modern biomedical research. Publicly available data have previously been provided in a broad array of diverse formats, making access to this very difficult. The publication and wide implementation of the Human Proteome Organisation Proteomics Standards Initiative Molecular Interactions (HUPO PSI-MI) format in 2004 was a major step towards the establishment of a single, unified format by which molecular interactions should be presented, but focused purely on protein-protein interactions. RESULTS: The HUPO-PSI has further developed the PSI-MI XML schema to enable the description of interactions between a wider range of molecular types, for example nucleic acids, chemical entities, and molecular complexes. Extensive details about each supported molecular interaction can now be captured, including the biological role of each molecule within that interaction, detailed description of interacting domains, and the kinetic parameters of the interaction. The format is supported by data management and analysis tools and has been adopted by major interaction data providers. Additionally, a simpler, tab-delimited format MITAB2.5 has been developed for the benefit of users who require only minimal information in an easy to access configuration. CONCLUSION: The PSI-MI XML2.5 and MITAB2.5 formats have been jointly developed by interaction data producers and providers from both the academic and commercial sector, and are already widely implemented and well supported by an active development community. PSI-MI XML2.5 enables the description of highly detailed molecular interaction data and facilitates data exchange between databases and users without loss of information. MITAB2.5 is a simpler format appropriate for fast Perl parsing or loading into Microsoft Excel.

Evaluation of the user-friendliness of seven new generation intensive care ventilators.

OBJECTIVE: To explore the user-friendliness and ergonomics of seven new generation intensive care ventilators. DESIGN: Prospective task-performing study. SETTING: Intensive care research laboratory, university hospital. METHODS: Ten physicians experienced in mechanical ventilation, but without prior knowledge of the ventilators, were asked to perform eight specific tasks [turning the ventilator on; recognizing mode and parameters; recognizing and setting alarms; mode change; finding and activating the pre-oxygenation function; pressure support setting; stand-by; finding and activating non-invasive ventilation (NIV) mode]. The time needed for each task was compared to a reference time (by trained physiotherapist familiar with the devices). A time >180 s was considered a task failure. RESULTS: For each of the tests on the ventilators, all physicians' times were significantly higher than the reference time (P < 0.001). A mean of 13 +/- 8 task failures (16%) was observed by the ventilator. The most frequently failed tasks were mode and parameter recognition, starting pressure support and finding the NIV mode. Least often failed tasks were turning on the pre-oxygenation function and alarm recognition and management. Overall, there was substantial heterogeneity between machines, some exhibiting better user-friendliness than others for certain tasks, but no ventilator was clearly better that the others on all points tested. CONCLUSIONS: The present study adds to the available literature outlining the ergonomic shortcomings of mechanical ventilators. These results suggest that closer ties between end-users and manufacturers should be promoted, at an early development phase of these machines, based on the scientific evaluation of the cognitive processes involved by users in the clinical setting.

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.