Comparison of multi-year Ka-band propagation characteristics at four sites in North America and Europe

Tropospheric phenomena such as clouds and mainly rain cause higher attenuation at Ka-band than at lower frequencies. In this collaborative paper, the main results of four long-term Ka-band propagation campaigns are presented. The experiments are carried out in Ottawa, Canada (satellite Anik F2); Aveiro, Portugal; Madrid, Spain; and Toulouse, France (satellite HotBird 6 in the last three cases) and have been running since 2004 in Aveiro, 2006 in Ottawa and Madrid, and 2008 in Toulouse. After a brief introduction of the experiments, rain rate and excess attenuation results are discussed, first for a common two-year measurement period and later for the whole database available. Seasonal attenuation statistics for Madrid, Ottawa and Aveiro are compared. Finally, fade duration and fade slope statistics derived at three locations are presented and discussed.

New definitions of margin for multi-class Boosting algorithms

La familia de algoritmos de Boosting son un tipo de técnicas de clasificación y regresión que han demostrado ser muy eficaces en problemas de Visión Computacional. Tal es el caso de los problemas de detección, de seguimiento o bien de reconocimiento de caras, personas, objetos deformables y acciones. El primer y más popular algoritmo de Boosting, AdaBoost, fue concebido para problemas binarios. Desde entonces, muchas han sido las propuestas que han aparecido con objeto de trasladarlo a otros dominios más generales: multiclase, multilabel, con costes, etc. Nuestro interés se centra en extender AdaBoost al terreno de la clasificación multiclase, considerándolo como un primer paso para posteriores ampliaciones. En la presente tesis proponemos dos algoritmos de Boosting para problemas multiclase basados en nuevas derivaciones del concepto margen. El primero de ellos, PIBoost, está concebido para abordar el problema descomponiéndolo en subproblemas binarios. Por un lado, usamos una codificación vectorial para representar etiquetas y, por otro, utilizamos la función de pérdida exponencial multiclase para evaluar las respuestas. Esta codificación produce un conjunto de valores margen que conllevan un rango de penalizaciones en caso de fallo y recompensas en caso de acierto. La optimización iterativa del modelo genera un proceso de Boosting asimétrico cuyos costes dependen del número de etiquetas separadas por cada clasificador débil. De este modo nuestro algoritmo de Boosting tiene en cuenta el desbalanceo debido a las clases a la hora de construir el clasificador. El resultado es un método bien fundamentado que extiende de manera canónica al AdaBoost original. El segundo algoritmo propuesto, BAdaCost, está concebido para problemas multiclase dotados de una matriz de costes. Motivados por los escasos trabajos dedicados a generalizar AdaBoost al terreno multiclase con costes, hemos propuesto un nuevo concepto de margen que, a su vez, permite derivar una función de pérdida adecuada para evaluar costes. Consideramos nuestro algoritmo como la extensión más canónica de AdaBoost para este tipo de problemas, ya que generaliza a los algoritmos SAMME, Cost-Sensitive AdaBoost y PIBoost. Por otro lado, sugerimos un simple procedimiento para calcular matrices de coste adecuadas para mejorar el rendimiento de Boosting a la hora de abordar problemas estándar y problemas con datos desbalanceados. Una serie de experimentos nos sirven para demostrar la efectividad de ambos métodos frente a otros conocidos algoritmos de Boosting multiclase en sus respectivas áreas. En dichos experimentos se usan bases de datos de referencia en el área de Machine Learning, en primer lugar para minimizar errores y en segundo lugar para minimizar costes. Además, hemos podido aplicar BAdaCost con éxito a un proceso de segmentación, un caso particular de problema con datos desbalanceados. Concluimos justificando el horizonte de futuro que encierra el marco de trabajo que presentamos, tanto por su aplicabilidad como por su flexibilidad teórica. Abstract The family of Boosting algorithms represents a type of classification and regression approach that has shown to be very effective in Computer Vision problems. Such is the case of detection, tracking and recognition of faces, people, deformable objects and actions. The first and most popular algorithm, AdaBoost, was introduced in the context of binary classification. Since then, many works have been proposed to extend it to the more general multi-class, multi-label, costsensitive, etc... domains. Our interest is centered in extending AdaBoost to two problems in the multi-class field, considering it a first step for upcoming generalizations. In this dissertation we propose two Boosting algorithms for multi-class classification based on new generalizations of the concept of margin. The first of them, PIBoost, is conceived to tackle the multi-class problem by solving many binary sub-problems. We use a vectorial codification to represent class labels and a multi-class exponential loss function to evaluate classifier responses. This representation produces a set of margin values that provide a range of penalties for failures and rewards for successes. The stagewise optimization of this model introduces an asymmetric Boosting procedure whose costs depend on the number of classes separated by each weak-learner. In this way the Boosting procedure takes into account class imbalances when building the ensemble. The resulting algorithm is a well grounded method that canonically extends the original AdaBoost. The second algorithm proposed, BAdaCost, is conceived for multi-class problems endowed with a cost matrix. Motivated by the few cost-sensitive extensions of AdaBoost to the multi-class field, we propose a new margin that, in turn, yields a new loss function appropriate for evaluating costs. Since BAdaCost generalizes SAMME, Cost-Sensitive AdaBoost and PIBoost algorithms, we consider our algorithm as a canonical extension of AdaBoost to this kind of problems. We additionally suggest a simple procedure to compute cost matrices that improve the performance of Boosting in standard and unbalanced problems. A set of experiments is carried out to demonstrate the effectiveness of both methods against other relevant Boosting algorithms in their respective areas. In the experiments we resort to benchmark data sets used in the Machine Learning community, firstly for minimizing classification errors and secondly for minimizing costs. In addition, we successfully applied BAdaCost to a segmentation task, a particular problem in presence of imbalanced data. We conclude the thesis justifying the horizon of future improvements encompassed in our framework, due to its applicability and theoretical flexibility.

Some past decision theory models for euro construction model codes and for road drainage in Spain, and modest hints with multi-criteria methods for agro and environmental assurance

The authors are from UPM and are relatively grouped, and all have intervened in different academic or real cases on the subject, at different times as being of different age. With precedent from E. Torroja and A. Páez in Madrid Spain Safety Probabilistic models for concrete about 1957, now in ICOSSAR conferences, author J.M. Antón involved since autumn 1967 for euro-steel construction in CECM produced a math model for independent load superposition reductions, and using it a load coefficient pattern for codes in Rome Feb. 1969, practically adopted for European constructions, giving in JCSS Lisbon Feb. 1974 suggestion of union for concrete-steel-al.. That model uses model for loads like Gumbel type I, for 50 years for one type of load, reduced to 1 year to be added to other independent loads, the sum set in Gumbel theories to 50 years return period, there are parallel models. A complete reliability system was produced, including non linear effects as from buckling, phenomena considered somehow in actual Construction Eurocodes produced from Model Codes. The system was considered by author in CEB in presence of Hydraulic effects from rivers, floods, sea, in reference with actual practice. When redacting a Road Drainage Norm in MOPU Spain an optimization model was realized by authors giving a way to determine the figure of Return Period, 10 to 50 years, for the cases of hydraulic flows to be considered in road drainage. Satisfactory examples were a stream in SE of Spain with Gumbel Type I model and a paper of Ven Te Chow with Mississippi in Keokuk using Gumbel type II, and the model can be modernized with more varied extreme laws. In fact in the MOPU drainage norm the redacting commission acted also as expert to set a table of return periods for elements of road drainage, in fact as a multi-criteria complex decision system. These precedent ideas were used e.g. in wide Codes, indicated in symposia or meetings, but not published in journals in English, and a condensate of contributions of authors is presented. The authors are somehow involved in optimization for hydraulic and agro planning, and give modest hints of intended applications in presence of agro and environment planning as a selection of the criteria and utility functions involved in bayesian, multi-criteria or mixed decision systems. Modest consideration is made of changing in climate, and on the production and commercial systems, and on others as social and financial.

Pushover analysis for the seismic response prediction of cable-stayed bridges under multi-directional excitation

Cable-stayed bridges represent nowadays key points in transport networks and their seismic behavior needs to be fully understood, even beyond the elastic range of materials. Both nonlinear dynamic (NL-RHA) and static (pushover) procedures are currently available to face this challenge, each with intrinsic advantages and disadvantages, and their applicability in the study of the nonlinear seismic behavior of cable-stayed bridges is discussed here. The seismic response of a large number of finite element models with different span lengths, tower shapes and class of foundation soil is obtained with different procedures and compared. Several features of the original Modal Pushover Analysis (MPA) are modified in light of cable-stayed bridge characteristics, furthermore, an extension of MPA and a new coupled pushover analysis (CNSP) are suggested to estimate the complex inelastic response of such outstanding structures subjected to multi-axial strong ground motions.

Multi-microphone recordings (EigenMike32) inside a WFS system (Iosono) through Ambisonics techniques using Matlab

El audio multicanal ha avanzado a pasos agigantados en los últimos años, y no solo en las técnicas de reproducción, sino que en las de capitación también. Por eso en este proyecto se encuentran ambas cosas: un array microfónico, EigenMike32 de MH Acoustics, y un sistema de reproducción con tecnología Wave Field Synthesis, instalado Iosono en la Jade Höchscule Oldenburg. Para enlazar estos dos puntos de la cadena de audio se proponen dos tipos distintos de codificación: la reproducción de la toma horizontal del EigenMike32; y el 3er orden de Ambisonics (High Order Ambisonics, HOA), una técnica de codificación basada en Armónicos Esféricos mediante la cual se simula el campo acústico en vez de simular las distintas fuentes. Ambas se desarrollaron en el entorno Matlab y apoyadas por la colección de scripts de Isophonics llamada Spatial Audio Matlab Toolbox. Para probar éstas se llevaron a cabo una serie de test en los que se las comparó con las grabaciones realizadas a la vez con un Dummy Head, a la que se supone el método más aproximado a nuestro modo de escucha. Estas pruebas incluían otras grabaciones hechas con un Doble MS de Schoeps que se explican en el proyecto “Sally”. La forma de realizar éstas fue, una batería de 4 audios repetida 4 veces para cada una de las situaciones garbadas (una conversación, una clase, una calle y un comedor universitario). Los resultados fueron inesperados, ya que la codificación del tercer orden de HOA quedo por debajo de la valoración Buena, posiblemente debido a la introducción de material hecho para un array tridimensional dentro de uno de 2 dimensiones. Por el otro lado, la codificación que consistía en extraer los micrófonos del plano horizontal se mantuvo en el nivel de Buena en todas las situaciones. Se concluye que HOA debe seguir siendo probado con mayores conocimientos sobre Armónicos Esféricos; mientras que el otro codificador, mucho más sencillo, puede ser usado para situaciones sin mucha complejidad en cuanto a espacialidad. In the last years the multichannel audio has increased in leaps and bounds and not only in the playback techniques, but also in the recording ones. That is the reason of both things being in this project: a microphone array, EigenMike32 from MH Acoustics; and a playback system with Wave Field Synthesis technology, installed by Iosono in Jade Höchscule Oldenburg. To link these two points of the audio chain, 2 different kinds of codification are proposed: the reproduction of the EigenMike32´s horizontal take, and the Ambisonics´ third order (High Order Ambisonics, HOA), a codification technique based in Spherical Harmonics through which the acoustic field is simulated instead of the different sound sources. Both have been developed inside Matlab´s environment and supported by the Isophonics´ scripts collection called Spatial Audio Matlab Toolbox. To test these, a serial of tests were made in which they were compared with recordings made at the time by a Dummy Head, which is supposed to be the closest method to our hearing way. These tests included other recording and codifications made by a Double MS (DMS) from Schoeps which are explained in the project named “3D audio rendering through Ambisonics techniques: from multi-microphone recordings (DMS Schoeps) to a WFS system, through Matlab”. The way to perform the tests was, a collection made of 4 audios repeated 4 times for each recorded situation (a chat, a class, a street and college canteen or Mensa). The results were unexpected, because the HOA´s third order stood under the Well valuation, possibly caused by introducing material made for a tridimensional array inside one made only by 2 dimensions. On the other hand, the codification that consisted of extracting the horizontal plane microphones kept the Well valuation in all the situations. It is concluded that HOA should keep being tested with larger knowledge about Spherical Harmonics; while the other coder, quite simpler, can be used for situations without a lot of complexity with regards to spatiality.

Markov blanket-based approach for learning multi-dimensional Bayesian network classifiers: An application to predict the European Quality of Life-5 Dimensions (EQ-5D) from the 39-item Parkinson's Disease Questionnaire (PDQ-39)

Multi-dimensional Bayesian network classifiers (MBCs) are probabilistic graphical models recently proposed to deal with multi-dimensional classification problems, where each instance in the data set has to be assigned to more than one class variable. In this paper, we propose a Markov blanket-based approach for learning MBCs from data. Basically, it consists of determining the Markov blanket around each class variable using the HITON algorithm, then specifying the directionality over the MBC subgraphs. Our approach is applied to the prediction problem of the European Quality of Life-5 Dimensions (EQ-5D) from the 39-item Parkinson’s Disease Questionnaire (PDQ-39) in order to estimate the health-related quality of life of Parkinson’s patients. Fivefold cross-validation experiments were carried out on randomly generated synthetic data sets, Yeast data set, as well as on a real-world Parkinson’s disease data set containing 488 patients. The experimental study, including comparison with additional Bayesian network-based approaches, back propagation for multi-label learning, multi-label k-nearest neighbor, multinomial logistic regression, ordinary least squares, and censored least absolute deviations, shows encouraging results in terms of predictive accuracy as well as the identification of dependence relationships among class and feature variables.

Discernment of bee pollen loads using computer vision and one-class classification techniques

In this paper, we propose a system for authenticating local bee pollen against fraudulent samples using image processing and classification techniques. Our system is based on the colour properties of bee pollen loads and the use of one-class classifiers to reject unknown pollen samples. The latter classification techniques allow us to tackle the major difficulty of the problem, the existence of many possible fraudulent pollen types. Also presented is a multi-classifier model with an ambiguity discovery process to fuse the output of the one-class classifiers. The method is validated by authenticating Spanish bee pollen types, the overall accuracy of the final system of being 94%. Therefore, the system is able to rapidly reject the non-local pollen samples with inexpensive hardware and without the need to send the product to the laboratory.

Virtual Excursions: a new way to explore science in class

The educational platform Virtual Science Hub (ViSH) has been developed as part of the GLOBAL excursion European project. ViSH (http://vishub.org/) is a portal where teachers and scientist interact to create virtual excursions to science infrastructures. The main motivation behind the project was to connect teachers - and in consequence their students - to scientific institutions and their wide amount of infrastructures and resources they are working with. Thus the idea of a hub was born that would allow the two worlds of scientists and teachers to connect and to innovate science teaching. The core of the ViSH?s concept design is based on virtual excursions, which allow for a number of pedagogical models to be applied. According to our internal definition a virtual excursion is a tour through some digital context by teachers and pupils on a given topic that is attractive and has an educational purpose. Inquiry-based learning, project-based and problem-based learning are the most prominent approaches that a virtual excursion may serve. The domain specific resources and scientific infrastructures currently available on the ViSH are focusing on life sciences, nano-technology, biotechnology, grid and volunteer computing. The virtual excursion approach allows an easy combination of these resources into interdisciplinary teaching scenarios. In addition, social networking features support the users in collaborating and communicating in relation to these excursions and thus create a community of interest for innovative science teaching. The design and development phases were performed following a participatory design approach. An important aspect in this process was to create design partnerships amongst all actors involved, researchers, developers, infrastructure providers, teachers, social scientists, and pedagogical experts early in the project. A joint sense of ownership was created and important changes during the conceptual phase were implemented in the ViSH due to early user feedback. Technology-wise the ViSH is based on the latest web technologies in order to make it cross-platform compatible so that it works on several operative systems such as Windows, Mac or Linux and multi-device accessible, such as desktop, tablet and mobile devices. The platform has been developed in HTML5, the latest standard for web development, assuring that it can run on any modern browser. In addition to social networking features a core element on the ViSH is the virtual excursions editor. It is a web tool that allows teachers and scientists to create rich mash-ups of learning resources provided by the e-Infrastructures (i.e. remote laboratories and live webcams). These rich mash-ups can be presented in either slides or flashcards format. Taking advantage of the web architecture supported, additional powerful components have been integrated like a recommendation engine to provide personalized suggestions about educational content or interesting users and a videoconference tool to enhance real-time collaboration like MashMeTV (http://www.mashme.tv/).

Multi-dimensional classification using Bayesian networks for stationary and evolving streaming data

Hoy en día, con la evolución continua y rápida de las tecnologías de la información y los dispositivos de computación, se recogen y almacenan continuamente grandes volúmenes de datos en distintos dominios y a través de diversas aplicaciones del mundo real. La extracción de conocimiento útil de una cantidad tan enorme de datos no se puede realizar habitualmente de forma manual, y requiere el uso de técnicas adecuadas de aprendizaje automático y de minería de datos. La clasificación es una de las técnicas más importantes que ha sido aplicada con éxito a varias áreas. En general, la clasificación se compone de dos pasos principales: en primer lugar, aprender un modelo de clasificación o clasificador a partir de un conjunto de datos de entrenamiento, y en segundo lugar, clasificar las nuevas instancias de datos utilizando el clasificador aprendido. La clasificación es supervisada cuando todas las etiquetas están presentes en los datos de entrenamiento (es decir, datos completamente etiquetados), semi-supervisada cuando sólo algunas etiquetas son conocidas (es decir, datos parcialmente etiquetados), y no supervisada cuando todas las etiquetas están ausentes en los datos de entrenamiento (es decir, datos no etiquetados). Además, aparte de esta taxonomía, el problema de clasificación se puede categorizar en unidimensional o multidimensional en función del número de variables clase, una o más, respectivamente; o también puede ser categorizado en estacionario o cambiante con el tiempo en función de las características de los datos y de la tasa de cambio subyacente. A lo largo de esta tesis, tratamos el problema de clasificación desde tres perspectivas diferentes, a saber, clasificación supervisada multidimensional estacionaria, clasificación semisupervisada unidimensional cambiante con el tiempo, y clasificación supervisada multidimensional cambiante con el tiempo. Para llevar a cabo esta tarea, hemos usado básicamente los clasificadores Bayesianos como modelos. La primera contribución, dirigiéndose al problema de clasificación supervisada multidimensional estacionaria, se compone de dos nuevos métodos de aprendizaje de clasificadores Bayesianos multidimensionales a partir de datos estacionarios. Los métodos se proponen desde dos puntos de vista diferentes. El primer método, denominado CB-MBC, se basa en una estrategia de envoltura de selección de variables que es voraz y hacia delante, mientras que el segundo, denominado MB-MBC, es una estrategia de filtrado de variables con una aproximación basada en restricciones y en el manto de Markov. Ambos métodos han sido aplicados a dos problemas reales importantes, a saber, la predicción de los inhibidores de la transcriptasa inversa y de la proteasa para el problema de infección por el virus de la inmunodeficiencia humana tipo 1 (HIV-1), y la predicción del European Quality of Life-5 Dimensions (EQ-5D) a partir de los cuestionarios de la enfermedad de Parkinson con 39 ítems (PDQ-39). El estudio experimental incluye comparaciones de CB-MBC y MB-MBC con los métodos del estado del arte de la clasificación multidimensional, así como con métodos comúnmente utilizados para resolver el problema de predicción de la enfermedad de Parkinson, a saber, la regresión logística multinomial, mínimos cuadrados ordinarios, y mínimas desviaciones absolutas censuradas. En ambas aplicaciones, los resultados han sido prometedores con respecto a la precisión de la clasificación, así como en relación al análisis de las estructuras gráficas que identifican interacciones conocidas y novedosas entre las variables. La segunda contribución, referida al problema de clasificación semi-supervisada unidimensional cambiante con el tiempo, consiste en un método nuevo (CPL-DS) para clasificar flujos de datos parcialmente etiquetados. Los flujos de datos difieren de los conjuntos de datos estacionarios en su proceso de generación muy rápido y en su aspecto de cambio de concepto. Es decir, los conceptos aprendidos y/o la distribución subyacente están probablemente cambiando y evolucionando en el tiempo, lo que hace que el modelo de clasificación actual sea obsoleto y deba ser actualizado. CPL-DS utiliza la divergencia de Kullback-Leibler y el método de bootstrapping para cuantificar y detectar tres tipos posibles de cambio: en las predictoras, en la a posteriori de la clase o en ambas. Después, si se detecta cualquier cambio, un nuevo modelo de clasificación se aprende usando el algoritmo EM; si no, el modelo de clasificación actual se mantiene sin modificaciones. CPL-DS es general, ya que puede ser aplicado a varios modelos de clasificación. Usando dos modelos diferentes, el clasificador naive Bayes y la regresión logística, CPL-DS se ha probado con flujos de datos sintéticos y también se ha aplicado al problema real de la detección de código malware, en el cual los nuevos ficheros recibidos deben ser continuamente clasificados en malware o goodware. Los resultados experimentales muestran que nuestro método es efectivo para la detección de diferentes tipos de cambio a partir de los flujos de datos parcialmente etiquetados y también tiene una buena precisión de la clasificación. Finalmente, la tercera contribución, sobre el problema de clasificación supervisada multidimensional cambiante con el tiempo, consiste en dos métodos adaptativos, a saber, Locally Adpative-MB-MBC (LA-MB-MBC) y Globally Adpative-MB-MBC (GA-MB-MBC). Ambos métodos monitorizan el cambio de concepto a lo largo del tiempo utilizando la log-verosimilitud media como métrica y el test de Page-Hinkley. Luego, si se detecta un cambio de concepto, LA-MB-MBC adapta el actual clasificador Bayesiano multidimensional localmente alrededor de cada nodo cambiado, mientras que GA-MB-MBC aprende un nuevo clasificador Bayesiano multidimensional. El estudio experimental realizado usando flujos de datos sintéticos multidimensionales indica los méritos de los métodos adaptativos propuestos. ABSTRACT Nowadays, with the ongoing and rapid evolution of information technology and computing devices, large volumes of data are continuously collected and stored in different domains and through various real-world applications. Extracting useful knowledge from such a huge amount of data usually cannot be performed manually, and requires the use of adequate machine learning and data mining techniques. Classification is one of the most important techniques that has been successfully applied to several areas. Roughly speaking, classification consists of two main steps: first, learn a classification model or classifier from an available training data, and secondly, classify the new incoming unseen data instances using the learned classifier. Classification is supervised when the whole class values are present in the training data (i.e., fully labeled data), semi-supervised when only some class values are known (i.e., partially labeled data), and unsupervised when the whole class values are missing in the training data (i.e., unlabeled data). In addition, besides this taxonomy, the classification problem can be categorized into uni-dimensional or multi-dimensional depending on the number of class variables, one or more, respectively; or can be also categorized into stationary or streaming depending on the characteristics of the data and the rate of change underlying it. Through this thesis, we deal with the classification problem under three different settings, namely, supervised multi-dimensional stationary classification, semi-supervised unidimensional streaming classification, and supervised multi-dimensional streaming classification. To accomplish this task, we basically used Bayesian network classifiers as models. The first contribution, addressing the supervised multi-dimensional stationary classification problem, consists of two new methods for learning multi-dimensional Bayesian network classifiers from stationary data. They are proposed from two different points of view. The first method, named CB-MBC, is based on a wrapper greedy forward selection approach, while the second one, named MB-MBC, is a filter constraint-based approach based on Markov blankets. Both methods are applied to two important real-world problems, namely, the prediction of the human immunodeficiency virus type 1 (HIV-1) reverse transcriptase and protease inhibitors, and the prediction of the European Quality of Life-5 Dimensions (EQ-5D) from 39-item Parkinson’s Disease Questionnaire (PDQ-39). The experimental study includes comparisons of CB-MBC and MB-MBC against state-of-the-art multi-dimensional classification methods, as well as against commonly used methods for solving the Parkinson’s disease prediction problem, namely, multinomial logistic regression, ordinary least squares, and censored least absolute deviations. For both considered case studies, results are promising in terms of classification accuracy as well as regarding the analysis of the learned MBC graphical structures identifying known and novel interactions among variables. The second contribution, addressing the semi-supervised uni-dimensional streaming classification problem, consists of a novel method (CPL-DS) for classifying partially labeled data streams. Data streams differ from the stationary data sets by their highly rapid generation process and their concept-drifting aspect. That is, the learned concepts and/or the underlying distribution are likely changing and evolving over time, which makes the current classification model out-of-date requiring to be updated. CPL-DS uses the Kullback-Leibler divergence and bootstrapping method to quantify and detect three possible kinds of drift: feature, conditional or dual. Then, if any occurs, a new classification model is learned using the expectation-maximization algorithm; otherwise, the current classification model is kept unchanged. CPL-DS is general as it can be applied to several classification models. Using two different models, namely, naive Bayes classifier and logistic regression, CPL-DS is tested with synthetic data streams and applied to the real-world problem of malware detection, where the new received files should be continuously classified into malware or goodware. Experimental results show that our approach is effective for detecting different kinds of drift from partially labeled data streams, as well as having a good classification performance. Finally, the third contribution, addressing the supervised multi-dimensional streaming classification problem, consists of two adaptive methods, namely, Locally Adaptive-MB-MBC (LA-MB-MBC) and Globally Adaptive-MB-MBC (GA-MB-MBC). Both methods monitor the concept drift over time using the average log-likelihood score and the Page-Hinkley test. Then, if a drift is detected, LA-MB-MBC adapts the current multi-dimensional Bayesian network classifier locally around each changed node, whereas GA-MB-MBC learns a new multi-dimensional Bayesian network classifier from scratch. Experimental study carried out using synthetic multi-dimensional data streams shows the merits of both proposed adaptive methods.

Enhancing K-12 science education through a multi-device web tool to facilitate content integration and e-infrastructure access

An effective K-12 science education is essential to succeed in future phases of the curriculum and the e-Infrastructures for education provide new opportunities to enhance it. This paper presents ViSH Viewer, an innovative web tool to consume educational content which aims to facilitate e-Science infrastructures access through a next generation learning object called "Virtual Excursion". Virtual Excursions provide a new way to explore science in class by taking advantage of e-Infrastructure resources and their integration with other educational contents, resulting in the creation of a reusable, interoperable and granular learning object. In order to better understand how this tool can allow teachers and students a joyful exploration of e-Science, we also present three Virtual Excursion examples. Details about the design, development and the tool itself are explained in this paper as well as the concept, structure and metadata of the new learning object.

Multi-dimensional classification with super-classes

The multi-dimensional classification problem is a generalisation of the recently-popularised task of multi-label classification, where each data instance is associated with multiple class variables. There has been relatively little research carried out specific to multi-dimensional classification and, although one of the core goals is similar (modelling dependencies among classes), there are important differences; namely a higher number of possible classifications. In this paper we present method for multi-dimensional classification, drawing from the most relevant multi-label research, and combining it with important novel developments. Using a fast method to model the conditional dependence between class variables, we form super-class partitions and use them to build multi-dimensional learners, learning each super-class as an ordinary class, and thus explicitly modelling class dependencies. Additionally, we present a mechanism to deal with the many class values inherent to super-classes, and thus make learning efficient. To investigate the effectiveness of this approach we carry out an empirical evaluation on a range of multi-dimensional datasets, under different evaluation metrics, and in comparison with high-performing existing multi-dimensional approaches from the literature. Analysis of results shows that our approach offers important performance gains over competing methods, while also exhibiting tractable running time.

Assessing Creativity In Engineering Students: A Comparative Between Degrees and Sudents In First And Last Year

An online open access test (CREAX self-assessment) has been used in this work so that students from degrees in engineering in the Universidad Polite¿cnica of Madrid (UPM) could self-assess their creative competence after several classroom activities. Different groups from the first year course have been statistically compared using data from their assessment. These first year students had different professors in the subject ?Technical Drawing? and belonged to several degrees in the UPM. They were as well compared regarding sex and a group of first year students was also compared to another last year group of the degree so as to observe possible differences in the achievement of this competence. Only one difference was detected concerning sex in one of the degrees. Among degrees, the higher marks obtained by students who had done specific exercises for the development of creativity in class is highlighted. Finally, a significantly high mark was observed in students during their last year of degree with respect to first year students. The tool CREAX has become very useful in the assessment of this competence in the UPM degrees in which it has been implemented.

Efficient Monte Carlo methods for multi-dimensional learning with classifier chains

Multi-dimensional classification (MDC) is the supervised learning problem where an instance is associated with multiple classes, rather than with a single class, as in traditional classification problems. Since these classes are often strongly correlated, modeling the dependencies between them allows MDC methods to improve their performance – at the expense of an increased computational cost. In this paper we focus on the classifier chains (CC) approach for modeling dependencies, one of the most popular and highest-performing methods for multi-label classification (MLC), a particular case of MDC which involves only binary classes (i.e., labels). The original CC algorithm makes a greedy approximation, and is fast but tends to propagate errors along the chain. Here we present novel Monte Carlo schemes, both for finding a good chain sequence and performing efficient inference. Our algorithms remain tractable for high-dimensional data sets and obtain the best predictive performance across several real data sets.

A multi-scale analysis of transforming agricultural markets in the context of globalization: implications for natural resources, food prices and rural poverty in Latin America

Vivimos una época en la que el mundo se transforma aceleradamente. La globalización está siguiendo un curso imparable, la población mundial así como la población urbana siguen creciendo, y en los países emergentes los ingresos promedios aumentan, resultando en un cambio también acelerado de las dietas y hábitos alimentarios. En conjunto esos factores están causando un aumento fundamental de la demanda de alimentos. Junto con la apertura de los mercados agrícolas, estos procesos han provocado un crecimiento del comercio internacional de alimentos durante la última década. Dado que muchos países de América Latina están dotados de abundancia de recursos naturales, estas tendencias han producido un crecimiento rápido de las exportaciones de bienes primarios desde América Latina al resto del mundo. En sólo 30 años la participación en el mercado agrícola de América Latina casi se ha duplicado, desde 10% en 1980 a 18% en 2010. Este aumento del comercio agrícola ha dado lugar a un debate sobre una serie de cuestiones cruciales relacionadas con los impactos del comercio en la seguridad alimentaria mundial, en el medio ambiente o en la reducción de la pobreza rural en países en desarrollo. Esta tesis aplica un marco integrado para analizar varios impactos relacionados con la transformación de los mercados agrícolas y los mercados rurales debidos a la globalización y, en particular, al progresivo aumento del comercio internacional. En concreto, la tesis aborda los siguientes temas: En primer lugar, la producción mundial de alimentos tendrá que aumentar considerablemente para poder satisfacer la demanda de una población mundial de 9000 millones personas en 2050, lo cual plantea grandes desafíos sobre los sistemas de la producción de alimentos. Alcanzar este logro, sin comprometer la integridad del medio ambiente en regiones exportadoras, es un reto aún mayor. En este contexto, la tesis analiza los efectos de la liberalización del comercio mundial, considerando distintas tecnologías de producción agraria, sobre unos indicadores de seguridad alimentaria en diferentes regiones del mundo y sobre distintos indicadores ambientales, teniendo en cuenta escalas diferentes en América Latina y el Caribe. La tesis utiliza el modelo “International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT)” – un modelo dinámico de equilibrio parcial del sector agrícola a escala global – para modelar la apertura de los mercados agrícolas así como diferentes escenarios de la producción hasta el año 2050. Los resultados del modelo están vinculados a modelos biofísicos para poder evaluar los cambios en la huella hídrica y la calidad del agua, así como para cuantificar los impactos del cambio en el uso del suelo sobre la biodiversidad y los stocks de carbono en 2050. Los resultados indican que la apertura de los mercados agrícolas es muy importante para mejorar la seguridad alimentaria a nivel mundial, sin embargo, produce también presiones ambientales indeseables en algunas regiones de América Latina. Contrastando dos escenarios que consideran distintas modos de producción, la expansión de la tierra agrícola frente a un escenario de la producción más intensiva, se demuestra que las mejoras de productividad son generalmente superiores a la expansión de las tierras agrícolas, desde un punto de vista económico e ambiental. En cambio, los escenarios de intensificación sostenible no sólo hacen posible una mayor producción de alimentos, sino que también generan menos impactos medioambientales que los otros escenarios futuros en todas sus dimensiones: biodiversidad, carbono, emisiones de nitratos y uso del agua. El análisis muestra que hay un “trade-off” entre el objetivo de alcanzar la sostenibilidad ambiental y el objetivo de la seguridad alimentaria, independiente del manejo agrícola en el futuro. En segundo lugar, a la luz de la reciente crisis de los precios de alimentos en los años 2007/08, la tesis analiza los impactos de la apertura de los mercados agrícolas en la transmisión de precios de los alimentos en seis países de América Latina: Argentina, Brasil, Chile, Colombia, México y el Perú. Para identificar las posibles relaciones de cointegración entre los índices de precios al consumidor de alimentos y los índices de precios de agrarios internacionales, sujetos a diferentes grados de apertura de mercados agrícolas en los seis países de América Latina, se utiliza un modelo simple de corrección de error (single equation error correction). Los resultados indican que la integración global de los mercados agrícolas ha dado lugar a diferentes tasas de transmisión de precios en los países investigados. Sobre todo en el corto plazo, las tasas de transmisión dependen del grado de apertura comercial, mientras que en el largo plazo las tasas de transmisión son elevadas, pero en gran medida independientes del régimen de comercio. Por lo tanto, durante un período de shocks de precios mundiales una mayor apertura del comercio trae consigo más inestabilidad de los precios domésticos a corto plazo y la resultante persistencia en el largo plazo. Sin embargo, estos resultados no verifican necesariamente la utilidad de las políticas comerciales, aplicadas frecuentemente por los gobiernos para amortiguar los shocks de precios. Primero, porque existe un riesgo considerable de volatilidad de los precios debido a cambios bruscos de la oferta nacional si se promueve la autosuficiencia en el país; y segundo, la política de proteccionismo asume el riesgo de excluir el país de participar en las cadenas de suministro de alto valor del sector agrícola, y por lo tanto esa política podría obstaculizar el desarrollo económico. Sin embargo, es indispensable establecer políticas efectivas para reducir la vulnerabilidad de los hogares a los aumentos repentinos de precios de alimentos, lo cual requiere una planificación gubernamental precisa con el presupuesto requerido disponible. En tercer lugar, la globalización afecta a la estructura de una economía y, por medios distintos, la distribución de los ingreso en un país. Perú sirve como ejemplo para investigar más profundamente las cuestiones relacionadas con los cambios en la distribución de los ingresos en zonas rurales. Perú, que es un país que está cada vez más integrado en los mercados mundiales, consiguió importantes descensos en la pobreza extrema en sus zonas rurales, pero a la vez adolece de alta incidencia de pobreza moderada y de desigualdad de los ingresos en zonas rural al menos durante el periodo comprendido entre 2004 y 2012. Esta parte de la tesis tiene como objetivo identificar las fuerzas impulsoras detrás de estas dinámicas en el Perú mediante el uso de un modelo de microsimulación basado en modelos de generación de ingresos aplicado a nivel los hogares rurales. Los resultados indican que la fuerza principal detrás de la reducción de la pobreza ha sido el crecimiento económico general de la economía, debido a las condiciones macroeconómicas favorables durante el periodo de estudio. Estos efectos de crecimiento beneficiaron a casi todos los sectores rurales, y dieron lugar a la disminución de la pobreza rural extrema, especialmente entre los agricultores de papas y de maíz. En parte, estos agricultores probablemente se beneficiaron de la apertura de los mercados agrícolas, que es lo que podría haber provocado un aumento de los precios al productor en tiempos de altos precios mundiales de los alimentos. Sin embargo, los resultados también sugieren que para una gran parte de la población más pobre existían barreras de entrada a la hora de poder participar en el empleo asalariado fuera de la agricultura o en la producción de cultivos de alto valor. Esto podría explicarse por la falta de acceso a unos activos importantes: por ejemplo, el nivel de educación de los pobres era apenas mejor en 2012 que en 2004; y también las dotaciones de tierra y de mano de obra, sobre todo de los productores pobres de maíz y patata, disminuyeron entre 2004 y 2012. Esto lleva a la conclusión de que aún hay margen para aplicar políticas para facilitar el acceso a estos activos, que podría contribuir a la erradicación de la pobreza rural. La tesis concluye que el comercio agrícola puede ser un importante medio para abastecer una población mundial creciente y más rica con una cantidad suficiente de calorías. Para evitar adversos efectos ambientales e impactos negativos para los consumidores y de los productores pobres, el enfoque debe centrarse en las mejoras de la productividad agrícola, teniendo en cuenta los límites ambientales y ser socialmente inclusivo. En este sentido, será indispensable seguir desarrollando soluciones tecnológicas que garanticen prácticas de producción agrícola minimizando el uso de recursos naturales. Además, para los pequeños pobres agricultores será fundamental eliminar las barreras de entrada a los mercados de exportación que podría tener efectos indirectos favorables a través de la adopción de nuevas tecnologías alcanzables a través de mercados internacionales. ABSTRACT The world is in a state of rapid transition. Ongoing globalization, population growth, rising living standards and increasing urbanization, accompanied by changing dietary patterns throughout the world, are increasing the demand for food. Together with more open trade regimes, this has triggered growing international agricultural trade during the last decade. For many Latin American countries, which are gifted with relative natural resource abundance, these trends have fueled rapid export growth of primary goods. In just 30 years, the Latin American agricultural market share has almost doubled from 10% in 1980 to 18% in 2010. These market developments have given rise to a debate around a number of crucial issues related to the role of agricultural trade for global food security, for the environment or for poverty reduction in developing countries. This thesis uses an integrated framework to analyze a broad array of possible impacts related to transforming agricultural and rural markets in light of globalization, and in particular of increasing trade activity. Specifically, the following issues are approached: First, global food production will have to rise substantially by the year 2050 to meet effective demand of a nine billion people world population which poses major challenges to food production systems. Doing so without compromising environmental integrity in exporting regions is an even greater challenge. In this context, the thesis explores the effects of future global trade liberalization on food security indicators in different world regions and on a variety of environmental indicators at different scales in Latin America and the Caribbean, in due consideration of different future agricultural production practices. The International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT) –a global dynamic partial equilibrium model of the agricultural sector developed by the International Food Policy Research Institute (IFPRI)– is applied to run different future production scenarios, and agricultural trade regimes out to 2050. Model results are linked to biophysical models, used to assess changes in water footprints and water quality, as well as impacts on biodiversity and carbon stocks from land use change by 2050. Results indicate that further trade liberalization is crucial for improving food security globally, but that it would also lead to more environmental pressures in some regions across Latin America. Contrasting land expansion versus more intensified agriculture shows that productivity improvements are generally superior to agricultural land expansion, from an economic and environmental point of view. Most promising for achieving food security and environmental goals, in equal measure, is the sustainable intensification scenario. However, the analysis shows that there are trade-offs between environmental and food security goals for all agricultural development paths. Second, in light of the recent food price crisis of 2007/08, the thesis looks at the impacts of increasing agricultural market integration on food price transmission from global to domestic markets in six Latin American countries, namely Argentina, Brazil, Chile, Colombia, Mexico and Peru. To identify possible cointegrating relationships between the domestic food consumer price indices and world food price levels, subject to different degrees of agricultural market integration in the six Latin American countries, a single equation error correction model is used. Results suggest that global agricultural market integration has led to different levels of price path-through in the studied countries. Especially in the short-run, transmission rates depend on the degree of trade openness, while in the long-run transmission rates are high, but largely independent of the country-specific trade regime. Hence, under world price shocks more trade openness brings with it more price instability in the short-term and the resulting persistence in the long-term. However, these findings do not necessarily verify the usefulness of trade policies, often applied by governments to buffer such price shocks. First, because there is a considerable risk of price volatility due to domestic supply shocks if self-sufficiency is promoted. Second, protectionism bears the risk of excluding a country from participating in beneficial high-value agricultural supply chains, thereby hampering economic development. Nevertheless, to reduce households’ vulnerability to sudden and large increases of food prices, effective policies to buffer food price shocks should be put in place, but must be carefully planned with the required budget readily available. Third, globalization affects the structure of an economy and, by different means, the distribution of income in a country. Peru serves as an example to dive deeper into questions related to changes in the income distribution in rural areas. Peru, a country being increasingly integrated into global food markets, experienced large drops in extreme rural poverty, but persistently high rates of moderate rural poverty and rural income inequality between 2004 and 2012. The thesis aims at disentangling the driving forces behind these dynamics by using a microsimulation model based on rural household income generation models. Results provide evidence that the main force behind poverty reduction was overall economic growth of the economy due to generally favorable macroeconomic market conditions. These growth effects benefited almost all rural sectors, and led to declines in extreme rural poverty, especially among potato and maize farmers. In part, these farmers probably benefited from policy changes towards more open trade regimes and the resulting higher producer prices in times of elevated global food price levels. However, the results also suggest that entry barriers existed for the poorer part of the population to participate in well-paid wage-employment outside of agriculture or in high-value crop production. This could be explained by a lack of sufficient access to important rural assets. For example, poor people’s educational attainment was hardly better in 2012 than in 2004. Also land and labor endowments, especially of (poor) maize and potato growers, rather decreased than increased over time. This leads to the conclusion that there is still scope for policy action to facilitate access to these assets, which could contribute to the eradication of rural poverty. The thesis concludes that agricultural trade can be one important means to provide a growing and richer world population with sufficient amounts of calories. To avoid adverse environmental effects and negative impacts for poor food consumers and producers, the focus should lie on agricultural productivity improvements, considering environmental limits and be socially inclusive. In this sense, it will be crucial to further develop technological solutions that guarantee resource-sparing agricultural production practices, and to remove entry barriers for small poor farmers to export markets which might allow for technological spill-over effects from high-value global agricultural supply chains.