Stacked Generalization (SG) is an ensemble learning technique, which aims to increase the performance of individual classifiers by combining them under a hierarchical architecture. This study consists of two major parts. In the first part, the performance of Stacked Generalization technique is analyzed with respect to the performance of the individual classifiers and the content of the training data. In the second part, based on the findings for a new class of algorithms, called Meta-Fuzzified Yield Value (Meta-FYV) is introduced. The first part introduces and verifies two hypotheses by a set of controlled experiments to assure the performance gain for SG. The learning mechanisms of SG to achieve high performance are explored and the relationship between the performance of the individual classifiers and that of SG is investigated. It is shown that if the samples in the training set are correctly classified by at least one base layer classifier, then, the generalization performance of the SG is increased, compared to the performance of the individual classifiers. In the second hypothesis, the effect of the spurious samples, which are not correctly labeled by any of the base layer classifiers, is investigated. In the second part of the thesis, six theorems are constructed based on the analysis of the feature spaces and the stacked generalization architecture. Based on the theorems and hypothesis, a new class of SG algorithms is proposed. The experiments are performed on both Corel data and synthetically generated data, using parallel programming techniques, on a high performance cluster.

We extend the study of learning and generalization in feed forward Boolean networks to random Boolean networks (RBNs). We explore the relationship between the learning capability and the network topology, the system size, the training sample size, and the complexity of the computational tasks. We show experimentally that there exists a critical connectivity Kc that improves the generalization and adaptation in networks. In addition, we show that in finite size networks, the critical K is a power-law function of the system size N and the fraction of inputs used during the training. We explain why adaptation improves at this critical connectivity by showing that the network ensemble manifests maximal topological diversity near Kc. Our work is partly motivated by self-assembled molecular and nanoscale electronics. Our findings allow to determine an automata network topology class for efficient and robust information processing.

Made available in DSpace on 2016-06-02T19:46:25Z (GMT). No. of bitstreams: 1 DissEMV.pdf: 791702 bytes, checksum: aec417790823990d5caeefebc08c7079 (MD5) Previous issue date: 2005-02-24 / Social Skills Training (SST) can be used to increase teachers social competence, contributing to the development of interactive contexts in the classroom as well as to the student s learning. The purpose of this research was to elaborate and describe intervention procedures used in a group SST program with teachers who work with children that present learning difficulties, consisting of 15 sessions for the development of social skills. The effects of the SST were analysed in terms of: a) acquisition of social skills classes and b) generalization of social skills learned in the sessions for the classroom context. The sample included 10 teachers, in the 26-54 age range. The data gathering procedure involved the use of the Interpersonal Relations Questionnaire (IRQ) before the intervention, the Social Skills Inventory (SSI) and the filming of each teacher with his/her students, before and after the intervention. The IRQ data showed that at the beginning the teachers had some difficulties in dealing with interpersonal conflicts in the classroom, demonstrating lack of social skills. It was also observed a certain incoherence in the verbal report related to some questions, highlighting flaws in the self-observation and self-knowledge repertoire. Based on the survey of the behavioral classes presented by the facilitator by means of the tape transcriptions, it was possible to classify the intervention procedures that had the following purposes: a) to investigate interpersonal difficulties; b) to analyse previous-consequent relations; c) to train specific social skills; d) presentation of positive results and; e) to sound out generalization which contributed to the learning of social skills and generalization for the classroom context. The SSI data showed that after the intervention there was an increase in the scores average of the group of teachers in all the social skills classes that were assessed. In the individual analysis some teachers showed much bigger changes than others and for some of them there was a slight decrease in the scores. As for the filming data in the classroom, there was no increase in all the social skills that were assessed, but in those that increased, the generalization was evidenced. The results obtained allow us to state that the improvement of the teachers social skills repertoire enables them to foster interactive contexts in the classroom, contributing to the student s academic and interpersonal development. / O Treinamento de Habilidades Sociais (THS) pode ser utilizado para aumentar a competência social de professores, contribuindo para o arranjo de contextos interativos em sala de aula e para a aprendizagem do aluno. O objetivo dessa pesquisa foi elaborar e descrever procedimentos de intervenção utilizados em um programa de THS em grupo, com professores de crianças com dificuldades de aprendizagem, composto por 15 sessões para o desenvolvimento de habilidades sociais. Analisaram-se os efeitos do THS em termos de: a) aquisição de classes de habilidades sociais; e b) generalização das habilidades sociais aprendidas nas sessões para o contexto de sala de aula. A amostra incluiu dez professores, de 26 a 54 anos. O procedimento de coleta de dados envolveu a aplicação do Questionário de Relações Interpessoais (QRI) antes da intervenção, do Inventário de Habilidades Sociais (IHS) e realização da filmagem de cada professora com seus alunos, pré e pós intervenção. Os dados do QRI mostraram que, inicialmente, as professoras apresentavam dificuldades de lidar com conflitos interpessoais em sala de aula, demonstrando um repertório deficitário de habilidades sociais. Observou-se, também, certa incoerência no relato verbal com relação a algumas questões, evidenciando falhas no repertório de auto-observação e de autoconhecimento. A partir do levantamento das classes comportamentais apresentadas pela facilitadora, por meio das transcrições de fitas cassetes, houve a possibilidade de classificar os procedimentos de intervenção que apresentavam os objetivos de: a) investigar dificuldades interpessoais; b) analisar relações antecedentes-conseqüentes; c) treinar habilidades sociais específicas; d) usar conseqüenciação positiva; e e) sondar generalização, os quais contribuíram para a aprendizagem das habilidades sociais e generalização para o contexto de sala de aula. Os dados do IHS mostraram que, após a intervenção, houve aumento nas médias dos escores do grupo de professoras em todas as classes de habilidades sociais avaliadas. Na análise individual, algumas professoras apresentaram mudanças bem maiores do que outras e, para algumas, houve uma pequena diminuição nos escores. Com relação aos dados da filmagem em sala de aula, não ocorreu aumento em todas as habilidades sociais avaliadas, mas naquelas que aumentaram, evidenciou-se generalização. Os resultados obtidos permitem afirmar que o aprimoramento do repertório de habilidades sociais de professores possibilita capacitá-los para a promoção de contextos interativos em sala de aula, contribuindo para o desenvolvimento acadêmico e interpessoal do aluno.

Using automatic methods to assign incoming tracks and albums from multiple sources to artists entities in a digital rights management company, where no universal artist identifier is available and artist names can be ambiguous, is a challenging problem. In this work we propose to use stacked generalization to combine the predictions of heterogeneous classifiers for an improved quality of artist content matching on two datasets from a digital rights management company. We compare the performance of using a nonlinear meta-level learner to a linear meta-level learner for the stacked generalization on the two datasets, as well as on eight additional datasets to see how well our results general- ize. We conduct experiments and evaluate how the different meta-level learners perform, using the base learners’ class probabilities or a combination of the base learners’ class probabilities and original input features as meta-features. Our results indicate that stacking with a non-linear meta-level learner can improve predictions on the artist chooser problem. Furthermore, our results indicate that when using a linear meta-level learner for stacked generalization, using the base learners’ class probabilities as metafeatures works best, while using a combination of the base learners’ class probabilities and the original input features as meta-features works best when using a non-linear metalevel learner. Among all the evaluated stacking approaches, stacking with a non-linear meta-level learner, using a combination of the base learners’ class probabilities and the original input features as meta-features, performs the best in our experiments over the ten evaluation datasets. / Att använda automatiska metoder för att tilldela spår och album från olika källor till artister i en digital underhållningstjänst är problematiskt då det inte finns några universellt använda identifierare för artister och namn på artister kan vara tvetydiga. I det här verket föreslår vi en användning av staplad generalisering för att kombinera förutsägningar från heterogena klassificerare för förbättra artistmatchningen i två datamäng från en digital underhållningstjänst. Vi jämför prestandan mellan en linjär och en icke-linjär metainlärningsmetod för den staplade generaliseringen av de två datamängder, samt även åtta ytterligare datamäng för att se hur resultaten kan generaliseras. Vi utför experiment och utvärderar hur de olika metainlärningsmetoderna presterar genom att använda basinlärningsmetodens klassannolikheter eller en kombination av basinlärningsmetodens klassannolikheter och den ursprungliga representationen som metarepresentation. Våra resultat indikerar att staplandet med en icke-linjär metainlärningsmetod kan förbättra förutsägningarna i problemet med att tilldela artister. Vidare indikerar våra resultat att när man använder en linjär metainlärningsmetod för en staplad generalisering är det bäst att använda basinlärningsmetodens klassannolikheter som metarepresentation, medan när man använder en icke-linjär metainlärningsmetod för en staplade generaliseringen är det bäst att använda en kombination av basinlärningsmetodens klassannolikheter och den ursprungliga representationen som metarepresentation. Av alla utvärderade sätt att stapla är staplandet med en icke-linjär metainlärningsmetod med en kombination av basinlärningsmetodens klassannolikheter och den ursprungliga representationen som metarepresentation den ansats som presterar bäst i våra experiment över de tio datamängderna.