Joint Analysis of Multiple Algorithms and Performance Measures

There has been an increasing interest in the development of new methods using Pareto optimality to deal with multi-objective criteria (for example, accuracy and time complexity). Once one has developed an approach to a problem of interest, the problem is then how to compare it with the state of art. In machine learning, algorithms are typically evaluated by comparing their performance on different data sets by means of statistical tests. Standard tests used for this purpose are able to consider jointly neither performance measures nor multiple competitors at once. The aim of this paper is to resolve these issues by developing statistical procedures that are able to account for multiple competing measures at the same time and to compare multiple algorithms altogether. In particular, we develop two tests: a frequentist procedure based on the generalized likelihood-ratio test and a Bayesian procedure based on a multinomial-Dirichlet conjugate model. We further extend them by discovering conditional independences among measures to reduce the number of parameters of such models, as usually the number of studied cases is very reduced in such comparisons. Data from a comparison among general purpose classifiers is used to show a practical application of our tests.

Physician choice making and characteristics associated with using physician-rating websites: cross-sectional study

BACKGROUND: Over the past decade, physician-rating websites have been gaining attention in scientific literature and in the media. However, little knowledge is available about the awareness and the impact of using such sites on health care professionals. It also remains unclear what key predictors are associated with the knowledge and the use of physician-rating websites. OBJECTIVE: To estimate the current level of awareness and use of physician-rating websites in Germany and to determine their impact on physician choice making and the key predictors which are associated with the knowledge and the use of physician-rating websites. METHODS: This study was designed as a cross-sectional survey. An online panel was consulted in January 2013. A questionnaire was developed containing 28 questions; a pretest was carried out to assess the comprehension of the questionnaire. Several sociodemographic (eg, age, gender, health insurance status, Internet use) and 2 health-related independent variables (ie, health status and health care utilization) were included. Data were analyzed using descriptive statistics, chi-square tests, and t tests. Binary multivariate logistic regression models were performed for elaborating the characteristics of physician-rating website users. Results from the logistic regression are presented for both the observed and weighted sample. RESULTS: In total, 1505 respondents (mean age 43.73 years, SD 14.39; 857/1505, 57.25% female) completed our survey. Of all respondents, 32.09% (483/1505) heard of physician-rating websites and 25.32% (381/1505) already had used a website when searching for a physician. Furthermore, 11.03% (166/1505) had already posted a rating on a physician-rating website. Approximately 65.35% (249/381) consulted a particular physician based on the ratings shown on the websites; in contrast, 52.23% (199/381) had not consulted a particular physician because of the publicly reported ratings. Significantly higher likelihoods for being aware of the websites could be demonstrated for female participants (P<.001), those who were widowed (P=.01), covered by statutory health insurance (P=.02), and with higher health care utilization (P<.001). Health care utilization was significantly associated with all dependent variables in our multivariate logistic regression models (P<.001). Furthermore, significantly higher scores could be shown for health insurance status in the unweighted and Internet use in the weighted models. CONCLUSIONS: Neither health policy makers nor physicians should underestimate the influence of physician-rating websites. They already play an important role in providing information to help patients decide on an appropriate physician. Assuming there will be a rising level of public awareness, the influence of their use will increase well into the future. Future studies should assess the impact of physician-rating websites under experimental conditions and investigate whether physician-rating websites have the potential to reflect the quality of care offered by health care providers.

Does a peer model’s task proficiency influence children’s solution choice and innovation?

The current study investigated whether 4- to 6-year-old children’s task solution choice was influenced by the past proficiency of familiar peer models and the children’s personal prior task experience. Peer past proficiency was established through behavioral assessments of interactions with novel tasks alongside peer and teacher predictions of each child’s proficiency. Based on these assessments, one peer model with high past proficiency and one age-, sex-, dominance-, and popularity-matched peer model with lower past proficiency were trained to remove a capsule using alternative solutions from a three-solution artificial fruit task. Video demonstrations of the models were shown to children after they had either a personal successful interaction or no interaction with the task. In general, there was not a strong bias toward the high past-proficiency model, perhaps due to a motivation to acquire multiple methods and the salience of other transmission biases. However, there was some evidence of a model-based past-proficiency bias; when the high past-proficiency peer matched the participants’ original solution, there was increased use of that solution, whereas if the high past-proficiency peer demonstrated an alternative solution, there was increased use of the alternative social solution and novel solutions. Thus, model proficiency influenced innovation.

Tidal Energy ”Round Robin” Tests Comparisons between towing tank and circulating tank results

One key step of the industrial development of a tidal energy device is the testing of scale prototype devices within a controlled laboratory environment. At present, there is no available experimental protocol which addresses in a quantitative manner the differences which can be expected between results obtained from the different types of facilities currently employed for this type of testing. As a consequence, where differences between results are found it has been difficult to confirm the extent to which these differences relate to the device performance or to the test facility type. In the present study, a comparative ”Round Robin” testing programme has been conducted as part of the EC FP VII MaRINET program in order to evaluate the impact of different experimental facilities on the test results. The aim of the trials was to test the same model tidal turbine in four different test facilities to explore the sensitivity of the results to the choice of facility. The facilities comprised two towing tanks, of very different size, and two circulating water channels. Performance assessments in terms of torque, drag and inflow speed showed very similar results in all facilities. However, expected differences between the different tank types (circulating and towing) were observed in the fluctuations of torque and drag measurements. The main facility parameters which can influence the behaviour of the turbine were identified; in particular the effect of blockage was shown to be significant in cases yielding for high thrust coefficients, even at relatively small blockage ratios.

Mobility in vehicular networks with multiple access points to the infrastructure

Nowadays there is a huge evolution in the technological world and in the wireless networks. The electronic devices have more capabilities and resources over the years, which makes the users more and more demanding. The necessity of being connected to the global world leads to the arising of wireless access points in the cities to provide internet access to the people in order to keep the constant interaction with the world. Vehicular networks arise to support safety related applications and to improve the traffic flow in the roads; however, nowadays they are also used to provide entertainment to the users present in the vehicles. The best way to increase the utilization of the vehicular networks is to give to the users what they want: a constant connection to the internet. Despite of all the advances in the vehicular networks, there were several issues to be solved. The presence of dedicated infrastructure to vehicular networks is not wide yet, which leads to the need of using the available Wi-Fi hotspots and the cellular networks as access networks. In order to make all the management of the mobility process and to keep the user’s connection and session active, a mobility protocol is needed. Taking into account the huge number of access points present at the range of a vehicle for example in a city, it will be beneficial to take advantage of all available resources in order to improve all the vehicular network, either to the users and to the operators. The concept of multihoming allows to take advantage of all available resources with multiple simultaneous connections. This dissertation has as objectives the integration of a mobility protocol, the Network-Proxy Mobile IPv6 protocol, with a host-multihoming per packet solution in order to increase the performance of the network by using more resources simultaneously, the support of multi-hop communications, either in IPv6 or IPv4, the capability of providing internet access to the users of the network, and the integration of the developed protocol in the vehicular environment, with the WAVE, Wi-Fi and cellular technologies. The performed tests focused on the multihoming features implemented on this dissertation, and on the IPv4 network access for the normal users. The obtained results show that the multihoming addition to the mobility protocol improves the network performance and provides a better resource management. Also, the results show the correct operation of the developed protocol in a vehicular environment.

Dynamic Programming Approaches for Estimating and Applying Large-scale Discrete Choice Models

People go through their life making all kinds of decisions, and some of these decisions affect their demand for transportation, for example, their choices of where to live and where to work, how and when to travel and which route to take. Transport related choices are typically time dependent and characterized by large number of alternatives that can be spatially correlated. This thesis deals with models that can be used to analyze and predict discrete choices in large-scale networks. The proposed models and methods are highly relevant for, but not limited to, transport applications. We model decisions as sequences of choices within the dynamic discrete choice framework, also known as parametric Markov decision processes. Such models are known to be difficult to estimate and to apply to make predictions because dynamic programming problems need to be solved in order to compute choice probabilities. In this thesis we show that it is possible to explore the network structure and the flexibility of dynamic programming so that the dynamic discrete choice modeling approach is not only useful to model time dependent choices, but also makes it easier to model large-scale static choices. The thesis consists of seven articles containing a number of models and methods for estimating, applying and testing large-scale discrete choice models. In the following we group the contributions under three themes: route choice modeling, large-scale multivariate extreme value (MEV) model estimation and nonlinear optimization algorithms. Five articles are related to route choice modeling. We propose different dynamic discrete choice models that allow paths to be correlated based on the MEV and mixed logit models. The resulting route choice models become expensive to estimate and we deal with this challenge by proposing innovative methods that allow to reduce the estimation cost. For example, we propose a decomposition method that not only opens up for possibility of mixing, but also speeds up the estimation for simple logit models, which has implications also for traffic simulation. Moreover, we compare the utility maximization and regret minimization decision rules, and we propose a misspecification test for logit-based route choice models. The second theme is related to the estimation of static discrete choice models with large choice sets. We establish that a class of MEV models can be reformulated as dynamic discrete choice models on the networks of correlation structures. These dynamic models can then be estimated quickly using dynamic programming techniques and an efficient nonlinear optimization algorithm. Finally, the third theme focuses on structured quasi-Newton techniques for estimating discrete choice models by maximum likelihood. We examine and adapt switching methods that can be easily integrated into usual optimization algorithms (line search and trust region) to accelerate the estimation process. The proposed dynamic discrete choice models and estimation methods can be used in various discrete choice applications. In the area of big data analytics, models that can deal with large choice sets and sequential choices are important. Our research can therefore be of interest in various demand analysis applications (predictive analytics) or can be integrated with optimization models (prescriptive analytics). Furthermore, our studies indicate the potential of dynamic programming techniques in this context, even for static models, which opens up a variety of future research directions.

Integración de dispositivos electrónicos inteligentes en Smart Grid

El sector eléctrico está experimentando cambios importantes tanto a nivel de gestión como a nivel de mercado. Una de las claves que están acelerando este cambio es la penetración cada vez mayor de los Sistemas de Generación Distribuida (DER), que están dando un mayor protagonismo al usuario a la hora de plantear la gestión del sistema eléctrico. La complejidad del escenario que se prevé en un futuro próximo, exige que los equipos de la red tenga la capacidad de interactuar en un sistema mucho más dinámico que en el presente, donde la interfaz de conexión deberá estar dotada de la inteligencia necesaria y capacidad de comunicación para que todo el sistema pueda ser gestionado en su conjunto de manera eficaz. En la actualidad estamos siendo testigos de la transición desde el modelo de sistema eléctrico tradicional hacia un nuevo sistema, activo e inteligente, que se conoce como Smart Grid. En esta tesis se presenta el estudio de un Dispositivo Electrónico Inteligente (IED) orientado a aportar soluciones para las necesidades que la evolución del sistema eléctrico requiere, que sea capaz de integrase en el equipamiento actual y futuro de la red, aportando funcionalidades y por tanto valor añadido a estos sistemas. Para situar las necesidades de estos IED se ha llevado a cabo un amplio estudio de antecedentes, comenzando por analizar la evolución histórica de estos sistemas, las características de la interconexión eléctrica que han de controlar, las diversas funciones y soluciones que deben aportar, llegando finalmente a una revisión del estado del arte actual. Dentro de estos antecedentes, también se lleva a cabo una revisión normativa, a nivel internacional y nacional, necesaria para situarse desde el punto de vista de los distintos requerimientos que deben cumplir estos dispositivos. A continuación se exponen las especificaciones y consideraciones necesarias para su diseño, así como su arquitectura multifuncional. En este punto del trabajo, se proponen algunos enfoques originales en el diseño, relacionados con la arquitectura del IED y cómo deben sincronizarse los datos, dependiendo de la naturaleza de los eventos y las distintas funcionalidades. El desarrollo del sistema continua con el diseño de los diferentes subsistemas que lo componen, donde se presentan algunos algoritmos novedosos, como el enfoque del sistema anti-islanding con detección múltiple ponderada. Diseñada la arquitectura y funciones del IED, se expone el desarrollo de un prototipo basado en una plataforma hardware. Para ello se analizan los requisitos necesarios que debe tener, y se justifica la elección de una plataforma embebida de altas prestaciones que incluye un procesador y una FPGA. El prototipo desarrollado se somete a un protocolo de pruebas de Clase A, según las normas IEC 61000-4-30 e IEC 62586-2, para comprobar la monitorización de parámetros. También se presentan diversas pruebas en las que se han estimado los retardos implicados en los algoritmos relacionados con las protecciones. Finalmente se comenta un escenario de prueba real, dentro del contexto de un proyecto del Plan Nacional de Investigación, donde este prototipo ha sido integrado en un inversor dotándole de la inteligencia necesaria para un futuro contexto Smart Grid.

Optimal choice of electoral preference data

Electoral researchers are so much accustomed to analyzing the choice of the single most preferred party as the left-hand side variable of their models of electoral behavior that they often ignore revealed preference data. Drawing on random utility theory, their models predict electoral behavior at the extensive margin of choice. Since the seminal work of Luce and others on individual choice behavior, however, many social science disciplines (consumer research, labor market research, travel demand, etc.) have extended their inventory of observed preference data with, for instance, multiple paired comparisons, complete or incomplete rankings, and multiple ratings. Eliciting (voter) preferences using these procedures and applying appropriate choice models is known to considerably increase the efficiency of estimates of causal factors in models of (electoral) behavior. In this paper, we demonstrate the efficiency gain when adding additional preference information to first preferences, up to full ranking data. We do so for multi-party systems of different sizes. We use simulation studies as well as empirical data from the 1972 German election study. Comparing the practical considerations for using ranking and single preference data results in suggestions for choice of measurement instruments in different multi-candidate and multi-party settings.

Trajectory design of a multiple flyby mission to asteroids

In designing the trajectory for a multiple flyby mission to asteroids the choice of the targets is the most challenging problem. This dissertation faces this problem in the framework of the recently issued medium-size mission call (M5) from ESA: CASTAway. Starting from the preliminary work done in [6], this thesis develops a methodology for sequencing the potential targets in a multiple flyby mission. In order to reduce the computational time, the complete database of known small bodies is firstly pruned on the base of heuristic considerations. Using the assumption of small manoeuvres, a chief orbit concept could be used. Thus, two heuristic thresholds are defined in order to exclude non-promising targets given a chief orbit. The sequencing process takes chief orbit and promising targets as inputs and gives a set of candidate sequences. The results of such a process are analysed in the CASTAway framework and the best feasible sequence studied in details.

Assessment of genetic diversity of zoonotic Brucella spp. recovered from livestock in Egypt using multiple locus VNTR analysis

Brucellosis is endemic in most parts of Egypt, where it is caused mainly by Brucella melitensis biovar 3, and affects cattle and small ruminants in spite of ongoing efforts devoted to its control. Knowledge of the predominant Brucella species/strains circulating in a region is a prerequisite of a brucellosis control strategy. For this reason a study aiming at the evaluation of the phenotypic and genetic heterogeneity of a panel of 17 Brucella spp. isolates recovered from domestic ruminants (cattle, buffalo, sheep, and goat) from four governorates during a period of five years (2002-2007) was carried out using microbiological tests and molecular biology techniques (PCR, MLVA-15, and sequencing). Thirteen strains were identified as B. melitensis biovar 3 while all phenotypic and genetic techniques classified the remaining isolates as B. abortus (n = 2) and B. suis biovar 1 (n = 2). MLVA-15 yielded a high discriminatory power (h = 0.801), indicating a high genetic diversity among the B. melitensis strains circulating among domestic ruminants in Egypt. This is the first report of the isolation of B. suis from cattle in Egypt which, coupled with the finding of B. abortus, suggests a potential role of livestock as reservoirs of several zoonotic Brucella species in the region.

Variability in the international normalised ratio (INR) in patients with antiphospholipid syndrome and positive lupus anticoagulant: should the INR targets be higher?

El síndrome antifosfolípido es un desorden autoinmune caracterizado por hipercoagulabilidad que requiere terapia anticoagulante como pilar fundamental, siendo la warfarina el tratamiento de elección en los casos que requieren manejo por largos periodos. Sin embargo, los pacientes con anticoagulante lúpico positivo representan un reto porque tienen mayor riesgo de presentar eventos trombóticos, sumado a que el seguimiento con el International Normalized Ratio (INR) no es confiable, ya que estos anticuerpos generan interferencia con las pruebas de laboratorio basadas en fosfolípidos, como es el caso del tiempo de protrombina (PT) con INR basal prolongado, incluso antes del inicio de la terapia anticoagulante. Por tal razón, se ilustra el caso de una paciente con síndrome antifosfolípido primario y anticoagulante lúpico positivo quien ha presentado múltiples episodios trombóticos, a pesar de recibir terapia anticoagulante. Además se hace una revisión de la literatura disponible y se postulan nuevas metas de INR en estos pacientes diferentes de las que se plantean actualmente.

Traffic Light Using Multiple Wireless Technologies

This paper presents the study and experimental tests for the viability analysis of using multiple wireless technologies in urban traffic light controllers in a Smart City environment. Communication drivers, different types of antennas, data acquisition methods and data processing for monitoring the network are presented. The sensors and actuators modules are connected in a local area network through two distinct low power wireless networks using both 868 MHz and 2.4 GHz frequency bands. All data communications using 868 MHz go through a Moteino. Various tests are made to assess the most advantageous features of each communication type. The experimental results show better range for 868 MHz solutions, whereas the 2.4 GHz presents the advantage of self-regenerating the network and mesh. The different pros and cons of both communication methods are presented.