Toxic compounds, such as pesticides, are routinely tested against a range of aquatic, avian and mammalian species as part of the registration process. The need for reducing dependence on animal testing has led to an increasing interest in alternative methods such as in silico modelling. The QSAR (Quantitative Structure Activity Relationship)-based models are already in use for predicting physicochemical properties, environmental fate, eco-toxicological effects, and specific biological endpoints for a wide range of chemicals. Data plays an important role in modelling QSARs and also in result analysis for toxicity testing processes. This research addresses number of issues in predictive toxicology. One issue is the problem of data quality. Although large amount of toxicity data is available from online sources, this data may contain some unreliable samples and may be defined as of low quality. Its presentation also might not be consistent throughout different sources and that makes the access, interpretation and comparison of the information difficult. To address this issue we started with detailed investigation and experimental work on DEMETRA data. The DEMETRA datasets have been produced by the EC-funded project DEMETRA. Based on the investigation, experiments and the results obtained, the author identified a number of data quality criteria in order to provide a solution for data evaluation in toxicology domain. An algorithm has also been proposed to assess data quality before modelling. Another issue considered in the thesis was the missing values in datasets for toxicology domain. Least Square Method for a paired dataset and Serial Correlation for single version dataset provided the solution for the problem in two different situations. A procedural algorithm using these two methods has been proposed in order to overcome the problem of missing values. Another issue we paid attention to in this thesis was modelling of multi-class data sets in which the severe imbalance class samples distribution exists. The imbalanced data affect the performance of classifiers during the classification process. We have shown that as long as we understand how class members are constructed in dimensional space in each cluster we can reform the distribution and provide more knowledge domain for the classifier.

As técnicas computacionais de aprendizagem de máquina para classificação ou categorização de dados estão sendo cada vez mais utilizadas no contexto de extração de informações ou padrões em bases de dados volumosas em variadas áreas de aplicação. Em paralelo, a aplicação destes métodos computacionais para identificação de padrões, bem como a classificação de dados relacionados ao consumo dos bens de informação é considerada uma tarefa complexa, visto que tais padrões de decisão do consumo estão relacionados com as preferências dos indivíduos e dependem de uma composição de características individuais, variáveis culturais, econômicas e sociais segregadas e agrupadas, além de ser um tópico pouco explorado no mercado brasileiro. Neste contexto, este trabalho realizou o estudo experimental a partir da aplicação do processo de Descoberta do conhecimento (KDD), o que inclui as etapas de seleção e Mineração de Dados, para um problema de classificação binária, indivíduos brasileiros que consomem e não consomem um bem de informação, filmes em salas de cinema, a partir dos dados obtidos na Pesquisa de Orçamento Familiar (POF) 2008-2009, pelo Instituto Brasileiro de Geografia e Estatística (IBGE). O estudo experimental resultou em uma análise comparativa da aplicação de duas técnicas de aprendizagem de máquina para classificação de dados, baseadas em aprendizado supervisionado, sendo estas Naïve Bayes (NB) e Support Vector Machine (SVM). Inicialmente, a revisão sistemática realizada com o objetivo de identificar estudos relacionados a aplicação de técnicas computacionais de aprendizado de máquina para classificação e identificação de padrões de consumo indica que a utilização destas técnicas neste contexto não é um tópico de pesquisa maduro e desenvolvido, visto que não foi abordado em nenhum dos trabalhos estudados. Os resultados obtidos a partir da análise comparativa realizada entre os algoritmos sugerem que a escolha dos algoritmos de aprendizagem de máquina para Classificação de Dados está diretamente relacionada a fatores como: (i) importância das classes para o problema a ser estudado; (ii) balanceamento entre as classes; (iii) universo de atributos a serem considerados em relação a quantidade e grau de importância destes para o classificador. Adicionalmente, os atributos selecionados pelo algoritmo de seleção de variáveis Information Gain sugerem que a decisão de consumo de cultura, mais especificamente do bem de informação, filmes em cinema, está fortemente relacionada a aspectos dos indivíduos relacionados a renda, nível de educação, bem como suas preferências por bens culturais / Machine learning techniques for data classification or categorization are increasingly being used for extracting information or patterns from volumous databases in various application areas. Simultaneously, the application of these computational methods to identify patterns, as well as data classification related to the consumption of information goods is considered a complex task, since such decision consumption paterns are related to the preferences of individuals and depend on a composition of individual characteristics, cultural, economic and social variables segregated and grouped, as well as being not a topic explored in the Brazilian market. In this context, this study performed an experimental study of application of the Knowledge Discovery (KDD) process, which includes data selection and data mining steps, for a binary classification problem, Brazilian individuals who consume and do not consume a information good, film at theaters in Brazil, from the microdata obtained from the Brazilian Household Budget Survey (POF), 2008-2009, performed by the Brazilian Institute of Geography and Statistics (IBGE). The experimental study resulted in a comparative analysis of the application of two machine-learning techniques for data classification, based on supervised learning, such as Naïve Bayes (NB) and Support Vector Machine (SVM). Initially, a systematic review with the objective of identifying studies related to the application of computational techniques of machine learning to classification and identification of consumption patterns indicates that the use of these techniques in this context is not a mature and developed research topic, since was not studied in any of the papers analyzed. The results obtained from the comparative analysis performed between the algorithms suggest that the choice of the machine learning algorithms for data classification is directly related to factors such as: (i) importance of the classes for the problem to be studied; (ii) balancing between classes; (iii) universe of attributes to be considered in relation to the quantity and degree of importance of these to the classifiers. In addition, the attributes selected by the Information Gain variable selection algorithm suggest that the decision to consume culture, more specifically information good, film at theaters, is directly related to aspects of individuals regarding income, educational level, as well as preferences for cultural goods

As técnicas computacionais de aprendizagem de máquina para classificação ou categorização de dados estão sendo cada vez mais utilizadas no contexto de extração de informações ou padrões em bases de dados volumosas em variadas áreas de aplicação. Em paralelo, a aplicação destes métodos computacionais para identificação de padrões, bem como a classificação de dados relacionados ao consumo dos bens de informação é considerada uma tarefa complexa, visto que tais padrões de decisão do consumo estão relacionados com as preferências dos indivíduos e dependem de uma composição de características individuais, variáveis culturais, econômicas e sociais segregadas e agrupadas, além de ser um tópico pouco explorado no mercado brasileiro. Neste contexto, este trabalho realizou o estudo experimental a partir da aplicação do processo de Descoberta do conhecimento (KDD), o que inclui as etapas de seleção e Mineração de Dados, para um problema de classificação binária, indivíduos brasileiros que consomem e não consomem um bem de informação, filmes em salas de cinema, a partir dos dados obtidos na Pesquisa de Orçamento Familiar (POF) 2008-2009, pelo Instituto Brasileiro de Geografia e Estatística (IBGE). O estudo experimental resultou em uma análise comparativa da aplicação de duas técnicas de aprendizagem de máquina para classificação de dados, baseadas em aprendizado supervisionado, sendo estas Naïve Bayes (NB) e Support Vector Machine (SVM). Inicialmente, a revisão sistemática realizada com o objetivo de identificar estudos relacionados a aplicação de técnicas computacionais de aprendizado de máquina para classificação e identificação de padrões de consumo indica que a utilização destas técnicas neste contexto não é um tópico de pesquisa maduro e desenvolvido, visto que não foi abordado em nenhum dos trabalhos estudados. Os resultados obtidos a partir da análise comparativa realizada entre os algoritmos sugerem que a escolha dos algoritmos de aprendizagem de máquina para Classificação de Dados está diretamente relacionada a fatores como: (i) importância das classes para o problema a ser estudado; (ii) balanceamento entre as classes; (iii) universo de atributos a serem considerados em relação a quantidade e grau de importância destes para o classificador. Adicionalmente, os atributos selecionados pelo algoritmo de seleção de variáveis Information Gain sugerem que a decisão de consumo de cultura, mais especificamente do bem de informação, filmes em cinema, está fortemente relacionada a aspectos dos indivíduos relacionados a renda, nível de educação, bem como suas preferências por bens culturais / Machine learning techniques for data classification or categorization are increasingly being used for extracting information or patterns from volumous databases in various application areas. Simultaneously, the application of these computational methods to identify patterns, as well as data classification related to the consumption of information goods is considered a complex task, since such decision consumption paterns are related to the preferences of individuals and depend on a composition of individual characteristics, cultural, economic and social variables segregated and grouped, as well as being not a topic explored in the Brazilian market. In this context, this study performed an experimental study of application of the Knowledge Discovery (KDD) process, which includes data selection and data mining steps, for a binary classification problem, Brazilian individuals who consume and do not consume a information good, film at theaters in Brazil, from the microdata obtained from the Brazilian Household Budget Survey (POF), 2008-2009, performed by the Brazilian Institute of Geography and Statistics (IBGE). The experimental study resulted in a comparative analysis of the application of two machine-learning techniques for data classification, based on supervised learning, such as Naïve Bayes (NB) and Support Vector Machine (SVM). Initially, a systematic review with the objective of identifying studies related to the application of computational techniques of machine learning to classification and identification of consumption patterns indicates that the use of these techniques in this context is not a mature and developed research topic, since was not studied in any of the papers analyzed. The results obtained from the comparative analysis performed between the algorithms suggest that the choice of the machine learning algorithms for data classification is directly related to factors such as: (i) importance of the classes for the problem to be studied; (ii) balancing between classes; (iii) universe of attributes to be considered in relation to the quantity and degree of importance of these to the classifiers. In addition, the attributes selected by the Information Gain variable selection algorithm suggest that the decision to consume culture, more specifically information good, film at theaters, is directly related to aspects of individuals regarding income, educational level, as well as preferences for cultural goods

El texto completo de este trabajo no está disponible en el Repositorio Académico UPC por restricciones de la casa editorial donde ha sido publicado. / This paper proposes a technological solution using a predictive analysis model to identify and reduce the level of risk for type 2 diabetes mellitus (T2DM) through a wearable device. Our proposal is based on previous models that use the auto-classification algorithm together with the addition of new risk factors, which provide a greater contribution to the results of the presumptive diagnosis of the user who wants to check his level of risk. The purpose is the primary prevention of type 2 diabetes mellitus by a non-invasive method composed of the phases: (1) Capture and storage of risk factors; (2) Predictive analysis model; (3) Presumptive results and recommendations; and (4) Preventive treatment. The main contribution is in the development of the proposed application. / Revisión por pares

Studien implementerade ett hjärna-dator-gränssnitt med hjälp av EEG-instrumentet MindWave Mobile Headset. Vi undersökte om det var möjligt att utföra fyra operationer genom att använda tankemönster. Fyra försökspersoner deltog i studien. Deras uppgift var att tänka i två tankemönster i följd som resulterade i en operation. EEG-signalen förbehandlas så att en mönsterigenkänningsmetod (k-NN) lättare kunde urskilja två tankemönster ur signalen. Denna undersökning har till vår vetskap inte tidigare utförts och är därmed kunskapsluckan vi ämnar fylla. Att fylla denna kunskapslucka är av intresse för bland annat användargrupperna: rörelsehindrade, spelintresserade och Virtual Reality-användare. Vi tog fram en modell som modellerade det bästa möjliga utfallet av metodiken i föreliggande studie. Undersökningens resultat kunde inte användas för att göra slutsatser angående frågeställningen då detta skulle vara att post hoc-teoretisera. I modellen visades dock tre av fyra operationer vara genomförbara, med en indikation om att även den fjärde var möjlig att utföra. Resultatet indikerar att det finns anledning att utföra en fortsatt studie. Den föreslagna fortsatta studien bör innefatta nya mätningar som testas av modellen för att fullt ut besvara problemformuleringen. / This study implements a Brain-Computer-Interface using the EEG-instrument MindWave Mobile Headset. We studied the feasibility of performing four operations using thought patterns. Four test subjects participated in the study. Their task was to think in two subsequent thought patterns that resulted in an operation. The EEG-signal was pre-processed in such a way that a pattern recognition algorithm (k-NN) more easily could recognize two thought patterns in the signal. This study has to our knowledge not been done before and thus aims to fill this lack of knowledge in the scientific community. User groups that have an interest in filling this gap are, amongst others; disabled people, gamers, and Virtual Reality users. We created a model that modeled the best possible outcome of the method used in this study. Conclusions drawn from the result can not be used to fully answer the problem statement, since it would be to post hoc-theorize. However, three out of four operations were possible to perform in the model, with an indication that the fourth also was possible to perform. These results indicate that there are grounds to continue this study. The proposed continued study should include new measurements that are tested by the model to determine if it is feasible to distinguish all four operations.

En la actualidad, el rol que cumplen los bancos en la economía del país y el impacto que tienen en las diferentes clases sociales es cada vez más importante. Estos siempre han sido un mercado que históricamente ha recibido un gran número de quejas y reclamaciones. Es por ello que, un mal servicio por parte del proveedor, una deficiente calidad de los productos y un precio fuera de mercado son las principales razones por las que los clientes abandonan una entidad bancaria. Esta situación va aumentando cada vez más y los bancos muestran su preocupación por este problema intentando implementar modelos que hasta el momento no han logrado cumplir con los objetivos. Además, existe un elevado nivel de competencia que obliga a las entidades financieras a velar por la lealtad de sus clientes para intentar mantenerlos e incrementar su rentabilidad. Este proyecto propone un Modelo de Análisis Predictivo soportado con la herramienta SAP Predictive Analytics, con el fin de ayudar en la toma de decisiones para la retención o fidelización de los clientes potenciales con tendencia a la deserción en la entidad bancaria. Esta propuesta se realizó mediante la necesidad de los mismos de conocer la exactitud de deserción de sus clientes categorizados potenciales. Se desarrolló una interfaz web como canal entre el Modelo de Análisis Predictivo propuesto y la entidad bancaria, con el fin de mostrar el resultado obtenido por el modelo indicando la exactitud, en porcentaje, de los clientes con tendencia a desertar. Además, como Plan de continuidad se propone 2 proyectos en base a la escalabilidad del Modelo de análisis predictivo propuesto, apoyándonos en la información obtenida en la etapa de análisis del modelo mismo. / Currently, the role played by banks in the country's economy and the impact they have on different social classes is increasingly important. These have always been a market that has historically received a large number of complaints and claims. It is therefore, poor service by the supplier, poor product quality and a price outside the market are the main reasons why customers leave a bank. This situation is increasingly and banks. In addition, there is a high level of competition that forces financial institutions and the loyalty of their customers to try to maintain them and increase their profitability. This project proposes a Predictive Analysis Model supported with the SAP Predictive Analytics tool, in order to help in making decisions for the retention or loyalty of clients with the tendency to drop out in the bank. This proposal was made through the need to know the accuracy of the desertion of its categorized clients. A web interface is shown as a channel between the Predictive Analysis Model and the bank, in order to show the result by the model that indicates the accuracy, in percentage, of clients with a tendency to defect. In addition, as a continuity plan, 2 projects are proposed based on the scalability of the Predictive Analysis Model, based on the information in the analysis stage of the model itself. / Tesis

Une alternative prometteuse à la lutte chimique pour la régulation des ravageurs de culture consiste à favoriser les populations de leurs prédateurs en jouant sur la structure du paysage agricole. L'identification de structures spatio-temporelles favorables aux ennemis naturels peut se faire par l'exploration de scénarios paysagers via une modélisation couplée de paysages et de dynamiques de population. Dans cette approche, les dynamiques de populations sont simulées sur des paysages virtuels aux propriétés structurales contrôlées, et l'observation des motifs de populations associés permet l'identification de structures favorables. La modélisation des dynamiques de populations repose cependant sur une connaissance fine des processus écologiques et de leur variabilité entre les différentes unités du paysage. L'état actuel des connaissances sur les mécanismes écologiques régissant les dynamiques des ennemis naturels de la famille des carabidés demeure l'obstacle majeur à la recherche in silico de scénarios paysagers favorables. La littérature sur les liens entre motifs de population de carabes et variables paysagères permet de formuler un ensemble d'hypothèses en compétition sur ces mécanismes. Réduire le nombre de ces hypothèses en analysant les convergences entre les motifs de population qui leur sont associés, et étudier la stabilité de ces convergences le long d'un gradient paysager apparaît comme une première étape nécessaire vers l'amélioration de la connaissance sur les processus écologiques. Dans une première partie, nous proposons une heuristique méthodologique basée sur la simulation de modèles de réaction-diffusion porteurs de ces hypothèses en compétition. L'étude des motifs de population a permis d'effectuer une typologie des modèles en fonction de leur réponse à une variable paysagère, via un algorithme de classification, réduisant ainsi le nombre d’hypothèses en compétition. La sélection de l'hypothèse la plus plausible parmi cet ensemble irréductible doit s'effectuer sur la base d'une observation des motifs de population sur le terrain. Cela implique que ces derniers soient caractérisés à des résolutions spatiales et temporelles suffisantes pour sélectionner une unique hypothèse parmi celles en compétition. Dans la deuxième partie, nous proposons une heuristique méthodologique permettant de déterminer a priori des stratégies d'échantillonnage maximisant la robustesse de la sélection d'hypothèses écologiques. Dans un premier temps, la simulation de modèles de réaction-diffusion représentatifs des hypothèses écologiques en compétition permet de générer des données biologiques virtuelles en tout point de l'espace et du temps. Ces données biologiques sont ensuite échantillonnées suivant des protocoles différant dans l'effort total d'échantillonnage, le nombre de dates, le nombre de points par unité d'espace et le nombre de réplicats de paysages. Les motifs des populations sont caractérisés à partir de ces échantillons. Le potentiel des stratégies d'échantillonnage est évalué via un algorithme de classification qui classe les modèles biologiques selon les motifs de population associés. L'analyse des performances de classification, i.e. la capacité de l'algorithme à discriminer les processus écologiques, permet de sélectionner un protocole d'échantillonnage optimal. Nous montrons également que la manière de distribuer l'effort d'échantillonnage entre ses composantes spatiales et temporelles est un levier majeur sur l'inférence des processus écologiques. La réduction du nombre d'hypothèses en compétition et l'aide à l'échantillonnage pour la sélection de modèles répondent à un besoin fort dans le processus d'acquisition de connaissances écologiques pour l'exploration in silico de scénarios paysagers favorisant des services écosystémiques. Nous discutons dans une dernière partie des implications de nos travaux et de leurs perspectives d'amélioration. / A promising alternative to the chemical control of pests consists in favoring their natural enemies populations by managing the agricultural landscape structure. Identifying favorable spatio-temporal structures can be performed through the exploration of landscape scenarios using coupled models of landscapes and population dynamics. In this approach, population dynamics are simulated on virtual landscapes with controlled properties, and the observation of population patterns allows for the identification of favorable structures. Population modeling however relies on a good knowledge about the ecological processes and their variability within the landscape elements. Current state of knowledge about the ecological mechanisms underlying natural enemies’ of the carabid family population dynamics remains a major obstacle to in silico investigation of favorable landscape scenarios. Literature about the relationship between carabid population and landscape properties allows the formulation of competing hypotheses about these processes. Reducing the number of these hypotheses by analyzing the convergence between their associated population patterns and investigating the stability of their convergence along a landscape gradient appears to be a necessary tep towards a better knowledge about ecological processes. In a first step, we propose a heuristic method based on the simulation of reaction-diffusion models carrying these competing hypotheses. Comparing the population patterns allowed to set a model typology according to their response to the landscape variable, through a classification algorithm, thus reducing the initial number of competing hypotheses. The selection of the most likely hypothesis from this irreducible set must rely on the observation of population patterns on the field. This implies that population patterns are described with spatial and temporal resolutions that are fine enough to select a unique hypothesis among the ones in competition. In the second part, we propose a heuristic method that allows determining a priori sampling strategies that maximize the robustness of ecological hypotheses selection. The simulation of reaction-diffusion models carrying the ecological hypotheses allows to generate virtual population data in space and time. These data are then sampled using strategies differing in the total effort, number of sampling locations, dates and landscape replicates. Population patterns are described from these samples. The sampling strategies are assessed through a classification algorithm that classifies the models according to the associated patterns. The analysis of classification performances, i.e. the ability of the algorithm to discriminate the ecological processes, allows the selection of optimal sampling designs. We also show that the way the sampling effort is distributed between its spatial and temporal components is strongly impacting the ecological processes inference. Reducing the number of competing ecological hypotheses, along with the selection of sampling strategies for optimal model inference both meet a strong need in the process of knowledge improvement about the ecological processes for the exploration of landscape scenarios favoring ecosystem services. In the last chapter, we discuss the implications and future prospects of our work.

Atrial fibrillation is one of the most common cardiac rhythm disorders characterized by ever-increasing prevalence and incidence in the Czech Republic and abroad. The incidence of atrial fibrillation is reported at 2-4 % of the population, but due to the often asymptomatic course, the real prevalence is even higher. The aim of this work is to design an algorithm for automatic detection of atrial fibrillation in the ECG record. In the practical part of this work, an algorithm for the detection of atrial fibrillation is proposed. For the detection itself, the k-nearest neighbor method, the support vector method and the multilayer neural network were used to classify ECG signals using features indicating the variability of RR intervals and the presence of the P wave in the ECG recordings. The best detection was achieved by a model using a multilayer neural network classification with two hidden layers. Results of success indicators: Sensitivity 91.23 %, Specificity 99.20 %, PPV 91.23 %, F-measure 91.23 % and Accuracy 98.53 %.

<div>Intelligent machines, and more broadly, intelligent systems, are becoming increasingly common in the everyday lives of humans. Nonetheless, despite significant advancements in automation, human supervision and intervention are still essential in almost all sectors, ranging from manufacturing and transportation to disaster-management and healthcare. These intelligent machines<i> interact and collaborate</i> with humans in a way that demands a greater level of trust between human and machine. While a lack of trust can lead to a human's disuse of automation, over-trust can result in a human trusting a faulty autonomous system which could have negative consequences for the human. Therefore, human trust should be <i>calibrated </i>to optimize these human-machine interactions. This calibration can be achieved by designing human-aware automation that can infer human behavior and respond accordingly in real-time.</div><div><br></div><div>In this dissertation, I present a probabilistic framework to model and calibrate a human's trust and workload dynamics during his/her interaction with an intelligent decision-aid system. More specifically, I develop multiple quantitative models of human trust, ranging from a classical state-space model to a classification model based on machine learning techniques. Both models are parameterized using data collected through human-subject experiments. Thereafter, I present a probabilistic dynamic model to capture the dynamics of human trust along with human workload. This model is used to synthesize optimal control policies aimed at improving context-specific performance objectives that vary automation transparency based on human state estimation. I also analyze the coupled interactions between human trust and workload to strengthen the model framework. Finally, I validate the optimal control policies using closed-loop human subject experiments. The proposed framework provides a foundation toward widespread design and implementation of real-time adaptive automation based on human states for use in human-machine interactions.</div>

<p dir="ltr">The primary purpose of this thesis is to propose and demonstrate BioGRNsemble, a modular and flexible approach for inferencing gene regulatory networks from RNA-Seq data. Integrating the GENIE3 and GRNBoost2 algorithms, this ensembles-of-ensembles method attempts to balance the outputs of both models through averaging, before providing a trimmed-down gene regulatory network consisting of transcription and target genes. Using a Drosophila Eye Dataset, we were able to successfully test this novel methodology, and our validation analysis using an online database determined over 3500 gene links correctly detected, albeit out of almost 530,000 predictions, leaving plenty of room for improvement in the future.</p>