Mapping of Environmental Data Using Kernel-Based Methods

Recently, kernel-based Machine Learning methods have gained great popularity in many data analysis and data mining fields: pattern recognition, biocomputing, speech and vision, engineering, remote sensing etc. The paper describes the use of kernel methods to approach the processing of large datasets from environmental monitoring networks. Several typical problems of the environmental sciences and their solutions provided by kernel-based methods are considered: classification of categorical data (soil type classification), mapping of environmental and pollution continuous information (pollution of soil by radionuclides), mapping with auxiliary information (climatic data from Aral Sea region). The promising developments, such as automatic emergency hot spot detection and monitoring network optimization are discussed as well.

Incremental neural networks for function approximation

A new strategy for incremental building of multilayer feedforward neural networks is proposed in the context of approximation of functions from R-p to R-q using noisy data. A stopping criterion based on the properties of the noise is also proposed. Experimental results for both artificial and real data are performed and two alternatives of the proposed construction strategy are compared.

A protein interaction atlas for the nuclear receptors: properties and quality of a hub-based dimerisation network.

BACKGROUND: The nuclear receptors are a large family of eukaryotic transcription factors that constitute major pharmacological targets. They exert their combinatorial control through homotypic heterodimerisation. Elucidation of this dimerisation network is vital in order to understand the complex dynamics and potential cross-talk involved. RESULTS: Phylogeny, protein-protein interactions, protein-DNA interactions and gene expression data have been integrated to provide a comprehensive and up-to-date description of the topology and properties of the nuclear receptor interaction network in humans. We discriminate between DNA-binding and non-DNA-binding dimers, and provide a comprehensive interaction map, that identifies potential cross-talk between the various pathways of nuclear receptors. CONCLUSION: We infer that the topology of this network is hub-based, and much more connected than previously thought. The hub-based topology of the network and the wide tissue expression pattern of NRs create a highly competitive environment for the common heterodimerising partners. Furthermore, a significant number of negative feedback loops is present, with the hub protein SHP [NR0B2] playing a major role. We also compare the evolution, topology and properties of the nuclear receptor network with the hub-based dimerisation network of the bHLH transcription factors in order to identify both unique themes and ubiquitous properties in gene regulation. In terms of methodology, we conclude that such a comprehensive picture can only be assembled by semi-automated text-mining, manual curation and integration of data from various sources.

Testing Spatial Autocorrelation in Weighted Networks : the Modes Permutation Test

In a weighted spatial network, as specified by an exchange matrix, the variances of the spatial values are inversely proportional to the size of the regions. Spatial values are no more exchangeable under independence, thus weakening the rationale for ordinary permutation and bootstrap tests of spatial autocorrelation. We propose an alternative permutation test for spatial autocorrelation, based upon exchangeable spatial modes, constructed as linear orthogonal combinations of spatial values. The coefficients obtain as eigenvectors of the standardised exchange matrix appearing in spectral clustering, and generalise to the weighted case the concept of spatial filtering for connectivity matrices. Also, two proposals aimed at transforming an acessibility matrix into a exchange matrix with with a priori fixed margins are presented. Two examples (inter-regional migratory flows and binary adjacency networks) illustrate the formalism, rooted in the theory of spectral decomposition for reversible Markov chains.

Validation of network communicability metrics for the analysis of brain structural networks.

Computational network analysis provides new methods to analyze the brain's structural organization based on diffusion imaging tractography data. Networks are characterized by global and local metrics that have recently given promising insights into diagnosis and the further understanding of psychiatric and neurologic disorders. Most of these metrics are based on the idea that information in a network flows along the shortest paths. In contrast to this notion, communicability is a broader measure of connectivity which assumes that information could flow along all possible paths between two nodes. In our work, the features of network metrics related to communicability were explored for the first time in the healthy structural brain network. In addition, the sensitivity of such metrics was analysed using simulated lesions to specific nodes and network connections. Results showed advantages of communicability over conventional metrics in detecting densely connected nodes as well as subsets of nodes vulnerable to lesions. In addition, communicability centrality was shown to be widely affected by the lesions and the changes were negatively correlated with the distance from lesion site. In summary, our analysis suggests that communicability metrics that may provide an insight into the integrative properties of the structural brain network and that these metrics may be useful for the analysis of brain networks in the presence of lesions. Nevertheless, the interpretation of communicability is not straightforward; hence these metrics should be used as a supplement to the more standard connectivity network metrics.

Transition to parenthood: The role of social interaction and endogenous networks

Empirical studies indicate that the transition to parenthood is influenced by an individual's peer group. To study the mechanisms creating interdepen- dencies across individuals' transition to parenthood and its timing we apply an agent-based simulation model. We build a one-sex model and provide agents with three different characteristics regarding age, intended education and parity. Agents endogenously form their network based on social closeness. Network members then may influence the agents' transition to higher parity levels. Our numerical simulations indicate that accounting for social inter- actions can explain the shift of first-birth probabilities in Austria over the period 1984 to 2004. Moreover, we apply our model to forecast age-specific fertility rates up to 2016.

Plastic modifications within inhibitory control networks induced by practicing a stop-signal task: an electrical neuroimaging study.

INTRODUCTION: Inhibitory control refers to our ability to suppress ongoing motor, affective or cognitive processes and mostly depends on a fronto-basal brain network. Inhibitory control deficits participate in the emergence of several prominent psychiatric conditions, including attention deficit/hyperactivity disorder or addiction. The rehabilitation of these pathologies might therefore benefit from training-based behavioral interventions aiming at improving inhibitory control proficiency and normalizing the underlying neurophysiological mechanisms. The development of an efficient inhibitory control training regimen first requires determining the effects of practicing inhibition tasks. METHODS: We addressed this question by contrasting behavioral performance and electrical neuroimaging analyses of event-related potentials (ERPs) recorded from humans at the beginning versus the end of 1 h of practice on a stop-signal task (SST) involving the withholding of responses when a stop signal was presented during a speeded auditory discrimination task. RESULTS: Practicing a short SST improved behavioral performance. Electrophysiologically, ERPs differed topographically at 200 msec post-stimulus onset, indicative of the engagement of distinct brain network with learning. Source estimations localized this effect within the inferior frontal gyrus, the pre-supplementary motor area and the basal ganglia. CONCLUSION: Our collective results indicate that behavioral and brain responses during an inhibitory control task are subject to fast plastic changes and provide evidence that high-order fronto-basal executive networks can be modified by practicing a SST.

Kernel-based Mapping of Meteorological Fields in Complex Orography

Due to the advances in sensor networks and remote sensing technologies, the acquisition and storage rates of meteorological and climatological data increases every day and ask for novel and efficient processing algorithms. A fundamental problem of data analysis and modeling is the spatial prediction of meteorological variables in complex orography, which serves among others to extended climatological analyses, for the assimilation of data into numerical weather prediction models, for preparing inputs to hydrological models and for real time monitoring and short-term forecasting of weather.In this thesis, a new framework for spatial estimation is proposed by taking advantage of a class of algorithms emerging from the statistical learning theory. Nonparametric kernel-based methods for nonlinear data classification, regression and target detection, known as support vector machines (SVM), are adapted for mapping of meteorological variables in complex orography.With the advent of high resolution digital elevation models, the field of spatial prediction met new horizons. In fact, by exploiting image processing tools along with physical heuristics, an incredible number of terrain features which account for the topographic conditions at multiple spatial scales can be extracted. Such features are highly relevant for the mapping of meteorological variables because they control a considerable part of the spatial variability of meteorological fields in the complex Alpine orography. For instance, patterns of orographic rainfall, wind speed and cold air pools are known to be correlated with particular terrain forms, e.g. convex/concave surfaces and upwind sides of mountain slopes.Kernel-based methods are employed to learn the nonlinear statistical dependence which links the multidimensional space of geographical and topographic explanatory variables to the variable of interest, that is the wind speed as measured at the weather stations or the occurrence of orographic rainfall patterns as extracted from sequences of radar images. Compared to low dimensional models integrating only the geographical coordinates, the proposed framework opens a way to regionalize meteorological variables which are multidimensional in nature and rarely show spatial auto-correlation in the original space making the use of classical geostatistics tangled.The challenges which are explored during the thesis are manifolds. First, the complexity of models is optimized to impose appropriate smoothness properties and reduce the impact of noisy measurements. Secondly, a multiple kernel extension of SVM is considered to select the multiscale features which explain most of the spatial variability of wind speed. Then, SVM target detection methods are implemented to describe the orographic conditions which cause persistent and stationary rainfall patterns. Finally, the optimal splitting of the data is studied to estimate realistic performances and confidence intervals characterizing the uncertainty of predictions.The resulting maps of average wind speeds find applications within renewable resources assessment and opens a route to decrease the temporal scale of analysis to meet hydrological requirements. Furthermore, the maps depicting the susceptibility to orographic rainfall enhancement can be used to improve current radar-based quantitative precipitation estimation and forecasting systems and to generate stochastic ensembles of precipitation fields conditioned upon the orography.

Bayesian Networks and the Value of the Evidence for the Forensic Two-Trace Transfer Problem

Forensic scientists face increasingly complex inference problems for evaluating likelihood ratios (LRs) for an appropriate pair of propositions. Up to now, scientists and statisticians have derived LR formulae using an algebraic approach. However, this approach reaches its limits when addressing cases with an increasing number of variables and dependence relationships between these variables. In this study, we suggest using a graphical approach, based on the construction of Bayesian networks (BNs). We first construct a BN that captures the problem, and then deduce the expression for calculating the LR from this model to compare it with existing LR formulae. We illustrate this idea by applying it to the evaluation of an activity level LR in the context of the two-trace transfer problem. Our approach allows us to relax assumptions made in previous LR developments, produce a new LR formula for the two-trace transfer problem and generalize this scenario to n traces.

Local optima networks of hard combinatorial landscapes

Combinatorial optimization involves finding an optimal solution in a finite set of options; many everyday life problems are of this kind. However, the number of options grows exponentially with the size of the problem, such that an exhaustive search for the best solution is practically infeasible beyond a certain problem size. When efficient algorithms are not available, a practical approach to obtain an approximate solution to the problem at hand, is to start with an educated guess and gradually refine it until we have a good-enough solution. Roughly speaking, this is how local search heuristics work. These stochastic algorithms navigate the problem search space by iteratively turning the current solution into new candidate solutions, guiding the search towards better solutions. The search performance, therefore, depends on structural aspects of the search space, which in turn depend on the move operator being used to modify solutions. A common way to characterize the search space of a problem is through the study of its fitness landscape, a mathematical object comprising the space of all possible solutions, their value with respect to the optimization objective, and a relationship of neighborhood defined by the move operator. The landscape metaphor is used to explain the search dynamics as a sort of potential function. The concept is indeed similar to that of potential energy surfaces in physical chemistry. Borrowing ideas from that field, we propose to extend to combinatorial landscapes the notion of the inherent network formed by energy minima in energy landscapes. In our case, energy minima are the local optima of the combinatorial problem, and we explore several definitions for the network edges. At first, we perform an exhaustive sampling of local optima basins of attraction, and define weighted transitions between basins by accounting for all the possible ways of crossing the basins frontier via one random move. Then, we reduce the computational burden by only counting the chances of escaping a given basin via random kick moves that start at the local optimum. Finally, we approximate network edges from the search trajectory of simple search heuristics, mining the frequency and inter-arrival time with which the heuristic visits local optima. Through these methodologies, we build a weighted directed graph that provides a synthetic view of the whole landscape, and that we can characterize using the tools of complex networks science. We argue that the network characterization can advance our understanding of the structural and dynamical properties of hard combinatorial landscapes. We apply our approach to prototypical problems such as the Quadratic Assignment Problem, the NK model of rugged landscapes, and the Permutation Flow-shop Scheduling Problem. We show that some network metrics can differentiate problem classes, correlate with problem non-linearity, and predict problem hardness as measured from the performances of trajectory-based local search heuristics.

Implicit methods for qualitative modeling of gene regulatory networks.

Advancements in high-throughput technologies to measure increasingly complex biological phenomena at the genomic level are rapidly changing the face of biological research from the single-gene single-protein experimental approach to studying the behavior of a gene in the context of the entire genome (and proteome). This shift in research methodologies has resulted in a new field of network biology that deals with modeling cellular behavior in terms of network structures such as signaling pathways and gene regulatory networks. In these networks, different biological entities such as genes, proteins, and metabolites interact with each other, giving rise to a dynamical system. Even though there exists a mature field of dynamical systems theory to model such network structures, some technical challenges are unique to biology such as the inability to measure precise kinetic information on gene-gene or gene-protein interactions and the need to model increasingly large networks comprising thousands of nodes. These challenges have renewed interest in developing new computational techniques for modeling complex biological systems. This chapter presents a modeling framework based on Boolean algebra and finite-state machines that are reminiscent of the approach used for digital circuit synthesis and simulation in the field of very-large-scale integration (VLSI). The proposed formalism enables a common mathematical framework to develop computational techniques for modeling different aspects of the regulatory networks such as steady-state behavior, stochasticity, and gene perturbation experiments.

Probabilistic age classification with Bayesian networks: a study on the ossification status of the medial clavicular epiphysis

In the past few decades, the rise of criminal, civil and asylum cases involving young people lacking valid identification documents has generated an increase in the demand of age estimation. The chronological age or the probability that an individual is older or younger than a given age threshold are generally estimated by means of some statistical methods based on observations performed on specific physical attributes. Among these statistical methods, those developed in the Bayesian framework allow users to provide coherent and transparent assignments which fulfill forensic and medico-legal purposes. The application of the Bayesian approach is facilitated by using probabilistic graphical tools, such as Bayesian networks. The aim of this work is to test the performances of the Bayesian network for age estimation recently presented in scientific literature in classifying individuals as older or younger than 18 years of age. For these exploratory analyses, a sample related to the ossification status of the medial clavicular epiphysis available in scientific literature was used. Results obtained in the classification are promising: in the criminal context, the Bayesian network achieved, on the average, a rate of correct classifications of approximatively 97%, whilst in the civil context, the rate is, on the average, close to the 88%. These results encourage the continuation of the development and the testing of the method in order to support its practical application in casework.

Assessing the added value of wastewater-based epidemiology to monitor illicit drug use

Wastewater-based epidemiology consists in acquiring relevant information about the lifestyle and health status of the population through the analysis of wastewater samples collected at the influent of a wastewater treatment plant. Whilst being a very young discipline, it has experienced an astonishing development since its firs application in 2005. The possibility to gather community-wide information about drug use has been among the major field of application. The wide resonance of the first results sparked the interest of scientists from various disciplines. Since then, research has broadened in innumerable directions. Although being praised as a revolutionary approach, there was a need to critically assess its added value, with regard to the existing indicators used to monitor illicit drug use. The main, and explicit, objective of this research was to evaluate the added value of wastewater-based epidemiology with regards to two particular, although interconnected, dimensions of illicit drug use. The first is related to trying to understand the added value of the discipline from an epidemiological, or societal, perspective. In other terms, to evaluate if and how it completes our current vision about the extent of illicit drug use at the population level, and if it can guide the planning of future prevention measures and drug policies. The second dimension is the criminal one, with a particular focus on the networks which develop around the large demand in illicit drugs. The goal here was to assess if wastewater-based epidemiology, combined to indicators stemming from the epidemiological dimension, could provide additional clues about the structure of drug distribution networks and the size of their market. This research had also an implicit objective, which focused on initiating the path of wastewater- based epidemiology at the Ecole des Sciences Criminelles of the University of Lausanne. This consisted in gathering the necessary knowledge about the collection, preparation, and analysis of wastewater samples and, most importantly, to understand how to interpret the acquired data and produce useful information. In the first phase of this research, it was possible to determine that ammonium loads, measured directly in the wastewater stream, could be used to monitor the dynamics of the population served by the wastewater treatment plant. Furthermore, it was shown that on the long term, the population did not have a substantial impact on consumption patterns measured through wastewater analysis. Focussing on methadone, for which precise prescription data was available, it was possible to show that reliable consumption estimates could be obtained via wastewater analysis. This allowed to validate the selected sampling strategy, which was then used to monitor the consumption of heroin, through the measurement of morphine. The latter, in combination to prescription and sales data, provided estimates of heroin consumption in line with other indicators. These results, combined to epidemiological data, highlighted the good correspondence between measurements and expectations and, furthermore, suggested that the dark figure of heroin users evading harm-reduction programs, which would thus not be measured by conventional indicators, is likely limited. In the third part, which consisted in a collaborative study aiming at extensively investigating geographical differences in drug use, wastewater analysis was shown to be a useful complement to existing indicators. In particular for stigmatised drugs, such as cocaine and heroin, it allowed to decipher the complex picture derived from surveys and crime statistics. Globally, it provided relevant information to better understand the drug market, both from an epidemiological and repressive perspective. The fourth part focused on cannabis and on the potential of combining wastewater and survey data to overcome some of their respective limitations. Using a hierarchical inference model, it was possible to refine current estimates of cannabis prevalence in the metropolitan area of Lausanne. Wastewater results suggested that the actual prevalence is substantially higher compared to existing figures, thus supporting the common belief that surveys tend to underestimate cannabis use. Whilst being affected by several biases, the information collected through surveys allowed to overcome some of the limitations linked to the analysis of cannabis markers in wastewater (i.e., stability and limited excretion data). These findings highlighted the importance and utility of combining wastewater-based epidemiology to existing indicators about drug use. Similarly, the fifth part of the research was centred on assessing the potential uses of wastewater-based epidemiology from a law enforcement perspective. Through three concrete examples, it was shown that results from wastewater analysis can be used to produce highly relevant intelligence, allowing drug enforcement to assess the structure and operations of drug distribution networks and, ultimately, guide their decisions at the tactical and/or operational level. Finally, the potential to implement wastewater-based epidemiology to monitor the use of harmful, prohibited and counterfeit pharmaceuticals was illustrated through the analysis of sibutramine, and its urinary metabolite, in wastewater samples. The results of this research have highlighted that wastewater-based epidemiology is a useful and powerful approach with numerous scopes. Faced with the complexity of measuring a hidden phenomenon like illicit drug use, it is a major addition to the panoply of existing indicators. -- L'épidémiologie basée sur l'analyse des eaux usées (ou, selon sa définition anglaise, « wastewater-based epidemiology ») consiste en l'acquisition d'informations portant sur le mode de vie et l'état de santé d'une population via l'analyse d'échantillons d'eaux usées récoltés à l'entrée des stations d'épuration. Bien qu'il s'agisse d'une discipline récente, elle a vécu des développements importants depuis sa première mise en oeuvre en 2005, notamment dans le domaine de l'analyse des résidus de stupéfiants. Suite aux retombées médiatiques des premiers résultats de ces analyses de métabolites dans les eaux usées, de nombreux scientifiques provenant de différentes disciplines ont rejoint les rangs de cette nouvelle discipline en développant plusieurs axes de recherche distincts. Bien que reconnu pour son coté objectif et révolutionnaire, il était nécessaire d'évaluer sa valeur ajoutée en regard des indicateurs couramment utilisés pour mesurer la consommation de stupéfiants. En se focalisant sur deux dimensions spécifiques de la consommation de stupéfiants, l'objectif principal de cette recherche était focalisé sur l'évaluation de la valeur ajoutée de l'épidémiologie basée sur l'analyse des eaux usées. La première dimension abordée était celle épidémiologique ou sociétale. En d'autres termes, il s'agissait de comprendre si et comment l'analyse des eaux usées permettait de compléter la vision actuelle sur la problématique, ainsi que déterminer son utilité dans la planification des mesures préventives et des politiques en matière de stupéfiants actuelles et futures. La seconde dimension abordée était celle criminelle, en particulier, l'étude des réseaux qui se développent autour du trafic de produits stupéfiants. L'objectif était de déterminer si cette nouvelle approche combinée aux indicateurs conventionnels, fournissait de nouveaux indices quant à la structure et l'organisation des réseaux de distribution ainsi que sur les dimensions du marché. Cette recherche avait aussi un objectif implicite, développer et d'évaluer la mise en place de l'épidémiologie basée sur l'analyse des eaux usées. En particulier, il s'agissait d'acquérir les connaissances nécessaires quant à la manière de collecter, traiter et analyser des échantillons d'eaux usées, mais surtout, de comprendre comment interpréter les données afin d'en extraire les informations les plus pertinentes. Dans la première phase de cette recherche, il y pu être mis en évidence que les charges en ammonium, mesurées directement dans les eaux usées permettait de suivre la dynamique des mouvements de la population contributrice aux eaux usées de la station d'épuration de la zone étudiée. De plus, il a pu être démontré que, sur le long terme, les mouvements de la population n'avaient pas d'influence substantielle sur le pattern de consommation mesuré dans les eaux usées. En se focalisant sur la méthadone, une substance pour laquelle des données précises sur le nombre de prescriptions étaient disponibles, il a pu être démontré que des estimations exactes sur la consommation pouvaient être tirées de l'analyse des eaux usées. Ceci a permis de valider la stratégie d'échantillonnage adoptée, qui, par le bais de la morphine, a ensuite été utilisée pour suivre la consommation d'héroïne. Combinée aux données de vente et de prescription, l'analyse de la morphine a permis d'obtenir des estimations sur la consommation d'héroïne en accord avec des indicateurs conventionnels. Ces résultats, combinés aux données épidémiologiques ont permis de montrer une bonne adéquation entre les projections des deux approches et ainsi démontrer que le chiffre noir des consommateurs qui échappent aux mesures de réduction de risque, et qui ne seraient donc pas mesurés par ces indicateurs, est vraisemblablement limité. La troisième partie du travail a été réalisée dans le cadre d'une étude collaborative qui avait pour but d'investiguer la valeur ajoutée de l'analyse des eaux usées à mettre en évidence des différences géographiques dans la consommation de stupéfiants. En particulier pour des substances stigmatisées, telles la cocaïne et l'héroïne, l'approche a permis d'objectiver et de préciser la vision obtenue avec les indicateurs traditionnels du type sondages ou les statistiques policières. Globalement, l'analyse des eaux usées s'est montrée être un outil très utile pour mieux comprendre le marché des stupéfiants, à la fois sous l'angle épidémiologique et répressif. La quatrième partie du travail était focalisée sur la problématique du cannabis ainsi que sur le potentiel de combiner l'analyse des eaux usées aux données de sondage afin de surmonter, en partie, leurs limitations. En utilisant un modèle d'inférence hiérarchique, il a été possible d'affiner les actuelles estimations sur la prévalence de l'utilisation de cannabis dans la zone métropolitaine de la ville de Lausanne. Les résultats ont démontré que celle-ci est plus haute que ce que l'on s'attendait, confirmant ainsi l'hypothèse que les sondages ont tendance à sous-estimer la consommation de cannabis. Bien que biaisés, les données récoltées par les sondages ont permis de surmonter certaines des limitations liées à l'analyse des marqueurs du cannabis dans les eaux usées (i.e., stabilité et manque de données sur l'excrétion). Ces résultats mettent en évidence l'importance et l'utilité de combiner les résultats de l'analyse des eaux usées aux indicateurs existants. De la même façon, la cinquième partie du travail était centrée sur l'apport de l'analyse des eaux usées du point de vue de la police. Au travers de trois exemples, l'utilisation de l'indicateur pour produire du renseignement concernant la structure et les activités des réseaux de distribution de stupéfiants, ainsi que pour guider les choix stratégiques et opérationnels de la police, a été mise en évidence. Dans la dernière partie, la possibilité d'utiliser cette approche pour suivre la consommation de produits pharmaceutiques dangereux, interdits ou contrefaits, a été démontrée par l'analyse dans les eaux usées de la sibutramine et ses métabolites. Les résultats de cette recherche ont mis en évidence que l'épidémiologie par l'analyse des eaux usées est une approche pertinente et puissante, ayant de nombreux domaines d'application. Face à la complexité de mesurer un phénomène caché comme la consommation de stupéfiants, la valeur ajoutée de cette approche a ainsi pu être démontrée.