Reliability generalization (RG) is a method for meta-analysis of reliability coefficients to estimate average score reliability across studies, determine variation in reliability, and identify study-level moderator variables influencing score reliability. A total of 107 peer-reviewed RG studies published from 1998 to 2013 were systematically reviewed to characterize the meta-analytic methods employed and to evaluate quality of reporting practice against standards for transparency in meta-analysis reporting. Most commonly, RG studies meta-analyzed alpha coefficients, which were synthesized using an unweighted, fixed-effects model applied to untransformed coefficients. Moderator analyses most frequently included multiple regression and bivariate correlations employing a fixed-effects model on untransformed, unweighted coefficients. Based on a unit-weighted scoring system, mean reporting quality for RG studies was statistically less than that for a comparison study of 198 meta-analyses in the organizational sciences across 42 indicators; however, means were not statistically significantly different between the two studies when evaluating reporting quality on 18 indicators deemed essential to ethical reporting practice in meta-analyses. Since its inception a wide variety of statistical methods have been applied to RG, and meta-analysis of reliability coefficients has extended to fields outside of psychological measurement, such as medicine and business. A set of guidelines for conducting and reporting RG studies is provided.

The purpose of the study was to provide training to peer tutors to teach students with mild and moderate disabilities sales tax computation using a simultaneous prompting procedure with fidelity. Non-target information presented during training sessions and generalization to natural environment settings was assessed. A multiple probe (days) across participants design was used to evaluate the effectiveness of the intervention on the dependent variables. The results showed peer tutors could use the simultaneous prompting procedure with fidelity to teach sales tax computation and students with disabilities could acquire sales tax computation. The students acquired the non-target information presented and generalization occurred in a natural setting.

The Generalization of the Digital Terrain Model Based on the TIN Abstract This diploma thesis deals with the up to now way and the possibilities of the digital terrain model generalization based on the TIN (the triangulate irregular network). New suitable way of the generalization of the digital terrain model procured from laser scanning data is proposed on the base of the existing generalization methods designated for digital models. Laser scanning data is characterized by a high areal density so the basic requirement is computing speed, maintaining the terrain features, such as a ridge, valley, steep hill, saddle, depression … and so on. The proposed algorithm is compared with the results of suggested algorithms and results from the generalization by the geographic software, such as Atlas DMT and ArcGIS.

The aim of this work is to design automatic detection of spot heights from airborne laser scanning data and its generalization. The first part is about the idea of spot height: the importance of spot heights for altimetry representation on maps and suggestions concerning the scatter of spot heights on maps. In the following part, different approaches for terrain critical points extraction (peaks, depressions, passes) from digital terrain models are described as suitable places for spot heights location. Further, we introduce a description of geomorphological characteristics of these points by which is possible to identify its significance and generalize them. In accordance to analysis of Základní mapa ČR and recommendation for spot heights location described in literature we designed own method of automatic generation and generalization spot heights from digital terrain models. This method were implemented and there is a description of the test data and results on the real data DMR 5G of the algorithm in the end of the work. Powered by TCPDF (www.tcpdf.org)

The purpose of this dissertation was to determine the usefulness of the linguistic processes of Distortion, Deletion, Generalization, and Semantic Ill-Formedness as constructs which differentiate the verbal communication of families who express dissatisfaction with their current intrafamilial relationships from families expressing satisfaction with their current relationships. Specifically, it was hypothesized that dissatisfied families would use these linguistic structures to a greater extent in their interaction than would satisfied families. Thirty-one family triads (father, mother, and child) were obtained by asking families randomly selected from the local high school student directory to participate. The families were given a Revealed Differences questionnaire which they subsequently discussed together and a questionnaire regarding their satisfaction with their intrafamilial relationships. The discussions were recorded and transcribed. Each of 150 Surface Structures (a complete thought, usually a grammatical sentence) per family was scored for 11 subcategories of Distortion, Deletion, Generalization, and Semantic Ill-Formedness. Interrater reliabilities ranged from .86 to .98. A mean was computed for the questionnaire pertaining to satisfaction with family relationships. Six families who scored at least one half standard deviation below the mean comprised the "dissatisfied" family group, and six families who scored at least one half standard deviation above the mean comprised the "satisfied" family group. It was found that the dissatisfied families used significantly more Deletion (p The linguistic process of Deletion is theorized to result in impoverishing the speaker's model of the world and the behavioral choices available to the speaker. Similarily, the listener(s) who must respond to the impoverished model is limited in his response and behavioral options. Since all members of the dissatisfied families used this form of language, they perpetuate the impoverishing model of the world and the limitations on their behavior. It was concluded that, while not establishing an etiologic link between the use of Deletion and family dissatisfaction, Deletion is part of the current verbal interaction of families who express dissatisfaction. Further research involving families in which a member is symptomatic is warranted based on the findings of this study. Language may provide at least one form of explanation regarding the process by which families maintain homeostasis in the face of symptom development. The use of linguistic concepts shows promise as an intermediate link in family interaction theory as well as a form of intervention available to therapists.

<div>It is a central problem in both statistics and computer science to understand the theoretical foundation of machine learning, especially deep learning. During the past decade, deep learning has achieved remarkable successes in solving many complex artificial intelligence tasks. The aim of this dissertation is to understand deep neural networks (DNNs) and other nonparametric methods in machine learning. In particular, three machine learning models have been studied: weight normalized DNNs, sparse DNNs, and the compositional nonparametric model.</div><div></div><div><br></div><div>The first chapter presents a general framework for norm-based capacity control for <i>L_p,q</i> weight normalized DNNs. We establish the upper bound on the Rademacher complexities of this family. Especially, with an <i>L_1,infty</i> normalization, we discuss properties of a width-independent capacity control, which only depends on the depth by a square root term. Furthermore, if the activation functions are anti-symmetric, the bound on the Rademacher complexity is independent of both the width and the depth up to a log factor. In addition, we study the weight normalized deep neural networks with rectified linear units (ReLU) in terms of functional characterization and approximation properties. In particular, for an <i>L_1,infty</i> weight normalized network with ReLU, the approximation error can be controlled by the <i>L₁</i> norm of the output layer.</div><div></div><div><br></div><div>In the second chapter, we study <i>L_1,infty</i>-weight normalization for deep neural networks with bias neurons to achieve the sparse architecture. We theoretically establish the generalization error bounds for both regression and classification under the <i>L_1,infty</i>-weight normalization. It is shown that the upper bounds are independent of the network width and <i>k^1/2</i>-dependence on the network depth <i>k</i>. These results provide theoretical justifications on the usage of such weight normalization to reduce the generalization error. We also develop an easily implemented gradient projection descent algorithm to practically obtain a sparse neural network. We perform various experiments to validate our theory and demonstrate the effectiveness of the resulting approach.</div><div></div><div><br></div><div>In the third chapter, we propose a compositional nonparametric method in which a model is expressed as a labeled binary tree of <i>2k+1</i> nodes, where each node is either a summation, a multiplication, or the application of one of the <i>q</i> basis functions to one of the <i>m₁</i> covariates. We show that in order to recover a labeled binary tree from a given dataset, the sufficient number of samples is <i>O(k </i>log<i>(m₁q)+</i>log<i>(k!))</i>, and the necessary number of samples is <i>Omega(k </i>log<i>(m₁q)-</i>log<i>(k!))</i>. We further propose a greedy algorithm for regression in order to validate our theoretical findings through synthetic experiments.</div>

L’implémentation d’intention est une stratégie de planification de l’action qui se révèle être plus efficace que l’intention d’agir pour permettre d’atteindre des buts désirés. À la différence d’une intention qui ne spécifie que les comportements qu’on souhaite réaliser (“j’ai l’intention de faire Y”), la stratégie d’implémentation d’intention consiste à créer un lien entre une situation et un comportement dirigé vers l’atteinte du but dans un format langagier en “si…alors” (“si je rencontre la situation X, alors je réaliserai le comportement Y”). L’efficacité de cette stratégie réside dans sa capacité à créer des habitudes de façon instantanée, habitudes dont la littérature révèle la force à persister dans le temps. Dans ce travail de thèse, nous examinons l’efficacité de cette stratégie pour faire face à deux obstacles à la persistance des comportements que sont les coûts associés à la réalisation du comportement et les changements situationnels. Dans trois études reposant sur un paradigme de décision, nous avons fait varier le coût associé à la réalisation du comportement. Les résultats, bien que ténus, suggèrent que former une implémentation d’intention permet de maintenir la réalisation des comportements plus longtemps que de ne former qu’une intention. Dans trois études reposant sur une tâche de catégorisation, nous avons présenté des situations : similaires versus non-similaires à la situation X spécifiée dans l’implémentation d’intention, et affordant le comportement Y spécifié dans cette stratégie versus un autre comportement. Les résultats suggèrent qu’en comparaison de l’intention, l’implémentation d’intention permet davantage de généraliser la réalisation des comportements, c’est-à-dire que le comportement semble être réalisé dans d’autres situations que celle spécifiée. La généralisation ne s’observe toutefois pas pour tout type de situation, mais uniquement lorsque ces situations sont fortement physiquement similaires à la situation spécifiée et affordent le comportement spécifié. Nous discuterons de l’apport de différentes théories de psychologie sociale permettant de comprendre les effets observés. / Implementation intention is a planning strategy that promotes goal achievement beyond goal intentions. While goal intentions merely specify a desired goal-directed behavior (“I intend to do Y”), implementation intentions link a goal-relevant situation to a goal-directed behavior in an “if, then” format (“If I encounter the situation X, then I’ll perform the behavior Y”). The efficacy of this strategy to promote goal achievement originates from its ability to create instant habits, habits for which the literature reveal the strength to persist over time. In this work, we examine the efficacy of this strategy to shield ongoing goal pursuit from the cost of behaviors and the situational changes. In three studies using a decision paradigm, we varied the cost associated to the realization of the behavior. The results, even weak, suggest that forming an implementation intention leads to maintaining a response for a longer period of time than goal intentions. In three studies using a categorization paradigm, we presented situations: physically similar versus non-similar to the situation X specified in the implementation intention; that afforded the behavior Y specified in the plan versus a different behavior. The results suggest that, compared to forming goal intentions, forming implementation intentions leads to a generalization, meaning that the behavior is realized in other situations than the specified one. However, a generalization does not occur for all kind of situation but only when situations that are strongly physically similar to the specified one and that afford the critical behavior are encountered. We discuss the contribution of some social psychology theories to understand the observed results.

Children diagnosed with autism are often described as having limited or restricted activities that serve as reinforcers as compared to neurotypical peers. Many theories suggest that one of the many ways children develop is through participation in play. This results in children coming into contact with new environmental stimuli. The procedures used to enhance play skills for children diagnosed with autism typically involve training novel responses with novel stimuli (e.g., toys). This is often done using naturalistic procedures. Because multiple procedures are used, it is unclear what procedure or combination of procedures causes the increases in play repertoires. This study investigated an important component of the treatment package know as reciprocal imitation training. Specifically, the study examined whether increased opportunities, contingent imitation without the requirement to imitate, or contingent imitation with the requirement to imitate would increase the number of toys a child diagnosed with autism would play with. The results showed dramatic increases in the number of toys the child independently chose to play with and an increase in the spontaneous use of different response topographies across novel stimuli only when the student was required to imitate a model. The results are discussed in terms of mediated generalization, the use of common responses, stimulus class formation and stimulus class expansion.

Made available in DSpace on 2016-07-27T14:21:29Z (GMT). No. of bitstreams: 1 Kaliane Gomes Pio Rosa.pdf: 829572 bytes, checksum: eefdf549d1c3aec245af10e318e891c5 (MD5) Previous issue date: 2008-02-29 / The aim of the present research was to investigate a nine-year-old boy s behavior diagnosed as having autism. Behavior analysis strategies were used for the acquisition and generalization of the participant s behaviors. They were made through the outlining of a multiple baseline among behaviors followed by the generalization test. The selected behaviors to go through the intervention were: eye contact, right pronunciation and instruction follow-up. The intervention occurred in three conditions: by the researcher, by the mother and the researcher and by the mother alone. The results show that the procedures were effective in the acquisition and generalization of the behaviors trained by the participants. The results were discussed in terms of applied methodology, reached effects and data comparison with theory. After the research phase was finished, it was possible to state that behavioral changes that were shown by the participant were quite considerable and constitute resultant from the procedures to which they were subjected. / O objetivo do presente estudo foi intervir em três classes de comportamento de uma criança diagnosticada como autista do sexo masculino e com nove anos de idade. Para a aquisição e generalização das classes de comportamento foram utilizadas estratégias da análise do comportamento. Procedeu-se por meio de um delineamento de linha de base múltipla entre comportamentos seguidos pelo teste de generalização. Os repertórios comportamentais selecionados para sofrerem intervenção foram: contato ocular, pronúncia correta e seguir instrução. As intervenções ocorreram em três condições: pela pesquisadora, pela mãe e pesquisadora e somente pela mãe. Os resultados demonstraram que os procedimentos foram efetivos na aquisição e generalização dos comportamentos treinados. Os resultados foram discutidos em termos da metodologia aplicada, dos efeitos alcançados e da comparação dos dados com a teoria. Concluída a fase de pesquisa, pôde-se afirmar que as modificações comportamentais apresentadas pelo participante demonstraram que constituem resultantes dos procedimentos que o mesmo foi submetido.

Made available in DSpace on 2016-04-29T13:17:33Z (GMT). No. of bitstreams: 1 Mariana Chernicharo Guimaraes.pdf: 958742 bytes, checksum: a76380f7c23f492d42bd175d29cbf614 (MD5) Previous issue date: 2010-12-03 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The present study investigated if it was possible to a) teach manding topographies of asking through the procedure of fading out the verbal model; b) verify if generalization of the mands installed occurs to objects which were not directly taught; and c) verify if a progressive reinforcement schedule LAG (Lag 1, Lag 2 and Lag 3), in which the participant must emit n different responses from previous ones to produce reinforcement, produces varied responses. In a multiple baseline design, five manding topographies of asking were taught to one participant and four topographies to another. The teaching procedure of these mands consisted in the presentation of the objects, the objects were previously selected by a preference assessment, to the participant and first install ecoic asking responses, such as can I get I want give it to me can I borrow and can I play . Next, the experimenter faded out the verbal model until the response was emitted under control of the object presentation in the absence of the verbal model. During the teaching sessions some generalizations tests were conducted, in which new objects were presented and the participants should emit mands with these new objects. After the teaching sessions the procedure to produce varied manding responses started, with the Lag 1, Lag 2 and Lag 3 reinforcement schedule. The results indicate that it was possible to teach the asking topographies to both participants, the number of responses emitted in the absence of a verbal model increased during procedure, the ecoic responses decreased and other fading levels were used throughout the teaching procedure. Beside that, variability was produced by the reinforcement schedule Lag 1, Lag 2 and Lag 3, one participant when the Lag increases the number of topographies emitted increases also and the other participant there is a bigger diversity of topographies which biggest distribution of responses in the biggest LAG S value. Some new (not taught responses) and recombined (partially taught responses) responses occurred and those increased with the amplified Lag. The generalization of taught topographies occurred to new objects presented to participants / No presente estudo investigou-se se é possível, a) ensinar topografias de mando de solicitação por meio de um procedimento de fading do modelo verbal; b) verificar se ocorre generalização dos mandos instalados para objetos que não foram utilizados no ensino direto das topografias de mando, e c) verificar se um esquema de reforçamento progressivo LAG (Lag 1, Lag 2 e Lag 3), em que exige-se que o participante emita respostas diferentes das n anteriores para produzir o reforçamento, produz respostas variadas. Utilizando um delineamento de linha de base múltipla, foram ensinadas cinco topografias de mando de solicitação para um participante e quatro topografias para outro. O procedimento de ensino desses mandos consistiu em apresentar objetos, selecionados em testes de preferência, para o participante e inicialmente instalar respostas ecóicas de solicitação tais como Posso pegar... , Quero... , Dá pra mim... , Me empresta... e Posso brincar... . A seguir, o fading out do modelo ecóico apresentado pelo experimentador foi conduzido até que a resposta de solicitação fosse emitida sob controle do objeto apresentado na ausência do modelo verbal. Durante as sessões de ensino foram feitos alguns teses de generalização, em que novos objetos eram apresentados e os participantes deveriam emitir mandos com esses novos objetos. Após as sessões de ensino, começou o procedimento para gerar respostas variadas de mando que haviam sido instaladas nas sessões de ensino, utilizando do esquema de reforçamento Lag 1, Lag 2 e Lag 3. O resultado indica que foi possível o ensino das topografias de mando de solicitação para ambos os participantes, as respostas sem modelo ocorreram cada vez mais durante o procedimento, a utilização do nível ecóico foi diminuindo e outros níveis de fading foram usados ao longo do ensino, além disso, houve produção de variabilidade pelo esquema de reforçamento Lag 1, Lag 2 e Lag 3, para um dos participantes quando aumenta o Lag, aumenta o número de topografias emitidas e para o outro há uma maior diversidade de topografias com uma maior distribuição de respostas no maior valor de LAG. Ocorrem respostas novas (respostas que não foram ensinadas) e recombinadas (respostas que tinham partes de topografias que haviam sido ensinadas), que aumentam com o aumento do Lag. E que ocorreu a generalização das topografias ensinadas, para os novos objetos que foram apresentados