The general purpose of the current study was to evaluate the potency of different categories of reinforcers with young children diagnosed with developmental delays. The participants were two boys and one girl who were between the ages of seven and eight. In Phase 1, we evaluated the reinforcing potency of tokens, edible items, and leisure items by using a progressive ratio (PR) schedule. For two participants, we found that tokens resulted in the highest PR break points. For one participant, edibles resulted in the highest break points (tokens were found to have the lowest break points). In Phase 2, we evaluated the effects of presession access on the break points of edibles and tokens. This manipulation served as a preliminary analysis of the extent to which tokens might function as generalized conditioned reinforcers. During Phase 2, presession access altered the break points of edibles, but not tokens. The findings of the current study suggest that PR schedules may be useful as a means to better assess certain dimensions of tasks and how they affect reinforcer effectiveness (e.g., amount of effort the client is willing to exert, the duration at which the client willing to work, how many responses the client will emit, etc.), and to evaluate to what extent tokens actually function as generalized conditioned reinforcers.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第19747号 / 工博第4202号 / 新制||工||1648(附属図書館) / 32783 / 京都大学大学院工学研究科高分子化学専攻 / (主査)教授 辻井 敬亘, 教授 吉崎 武尚, 教授 渡辺 宏 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

<p dir="ltr">Reinforcement learning algorithms have traditionally been implemented with the goal</p><p dir="ltr">of maximizing a reward signal. By contrast, Decision Transformer (DT) uses a transformer</p><p dir="ltr">model to predict the next action in a sequence. The transformer model is trained on datasets</p><p dir="ltr">consisting of state, action, return trajectories. The original DT paper examined a small</p><p dir="ltr">number of environments, five from the Atari domain, and three from continuous control,</p><p dir="ltr">and one that examined credit assignment. While this gives an idea of what the decision</p><p dir="ltr">transformer can do, the variety of environments in the Atari domain are limited. In this</p><p dir="ltr">work, we propose an extension of the environments that decision transformer can be trained</p><p dir="ltr">on by adding support for the VizDoom environment. We also developed a faster method for</p><p dir="ltr">offline RL dataset generation, using Sample Factory, a library focused on high throughput,</p><p dir="ltr">to generate a dataset comparable in quality to existing methods using significantly less time.</p><p dir="ltr"><br></p>

An experimental program was conducted to investigate the strength and deformations of deep beams reinforced with Carbon Fibre Reinforced Polymer (CFRP) longitudinal and transverse reinforcement. Two groups of beams were tested, with each group comprising three beams. Two of the three beams in each group were reinforced with CFRP bars while the third beam was reinforced with conventional rebars and the latter beam was used as a control specimen. Beams in group 1 had span-to-depth ratio of one, while those in group 2 had a span-to-depth ratio of two. Beams in both groups had height of 900 mm and width of 250 mm. All the beams were simply supported and were tested in four-point bending with the point loads applied at one-third of the span. The test results revealed no significant difference between the behavior of the FRP reinforced beams and the companion control beams. On the other hand due to lack of hooks at the ends of the CFRP bars, and the loss of bond between the CFRP fibres and the sand grains on the surface of the bar, the failure in the CFRP reinforced beams was caused by the loss of anchorage while in the steel reinforced beams, the failure was initiated by the yielding of the longitudinal steel, followed by the crushing of the horizontal compression strut, but the nodal zones did not fail in any of the beams. Consequently, it was concluded that CFRP reinforced deep beams could be designed using the current CSA method for conventional steel reinforced concrete deep beams, provided the anchorage or bond strength of FRP bars could be properly determined. The existing nodal efficiency factors for the CCC nodal zones, as given in the CSA A23.3. standard, could be applied to CFRP reinforced beams while the corresponding factor for the CCT zone may be conservatively assumed to be 0.68. Finally, despite the linear elastic behavior of CFRP reinforcement, deep beams reinforced with CFRP bars could be designed using strut and tie models. / Thesis / Master of Applied Science (MASc)

Yes / A 9 x 18 x 1 feed-forward neural network (NN) model trained using a resilient back-propagation algorithm and early stopping technique is constructed to predict the shear strength of deep reinforced concrete beams. The input layer covering geometrical and material properties of deep beams has nine neurons, and the corresponding output is the shear strength. Training, validation and testing of the developed neural network have been achieved using a comprehensive database compiled from 362 simple and 71 continuous deep beam specimens. The shear strength predictions of deep beams obtained from the developed NN are in better agreement with test results than those determined from strut-and-tie models. The mean and standard deviation of the ratio between predicted capacities using the NN and measured shear capacities are 1.028 and 0.154, respectively, for simple deep beams, and 1.0 and 0.122, respectively, for continuous deep beams. In addition, the trends ascertained from parametric study using the developed NN have a consistent agreement with those observed in other experimental and analytical investigations.

Recent interest in utilizing hydrogen fuel cell technology for automotive applications has lead to concerns regarding the durability of fiber reinforced polymer (FRP) composite materials. Automotive fuel cell power train systems must prove themselves as a reliable alternative to the combustion engines and automatic transmissions. The use of polymer composites in fuel cells to serve as manifolds is promising because of their high strength to weight ratio, and they do not corrode like metals manifolds. Composite materials designed for use in Polymer Electrolyte Membrane (PEM) Fuel Cells are exposed to very high humidity environment and operated at elevated temperatures (~85°C). The susceptibility of fiber reinforced polymers to reduction in modulus, strength, and life in chemical environments has been well documented in the literature, especially at elevated temperatures. A chopped carbon fiber epoxy composite (Material A) and a chopped glass fiber epoxy composite (Material B) were exposed at 85°C to air, water, and a 50/50 water/antifreeze mixture, and periodically examined for tensile, compression, and flexural strengths at various temperatures. Following 2000 hours (83 days) of exposure, Materials A & B did not reach full saturation. Fatigue tests were conducted at various load levels and temperatures to determine their effect on cycles to failure, and carpet plots were generated. Blister formation in aged composites led to reductions in material properties as great as 25% to 75%. A mechanistic explanation was developed for the formation of blisters in the epoxy composite. Recommendations for material improvement and feasibility of material use for fuel cell manifolds and pressure plates were made. / Master of Science

Numerical analyses of two different geotechnical problems, namely a bridge abutment and a geosynthetic reinforced embankment are presented in this thesis. Settlement, bearing capacity and slope stability are the major factors that need to be considered in the design of a foundation near a sloped fill. In this thesis, the behaviour of a small scale model footing located near the shoulder of a sloped fill was investigated numerically. Single and multiple layers of geogrid were used to reinforce the sloped fill, and their effects on the load-deformation behaviour and bearing capacity of the footing were explored. The analyses showed 80%, 168%, 295% and 375% maximum improvement in the ultimate bearing capacity with 1, 2, 3 and 4 reinforcement layers respectively. This maximum improvement depends on the embedment depths of the reinforcement layers below the foundation and the suggested optimal depths are discussed. Typically, greatest improvement in ultimate bearing capacity with a single layer of reinforcement was obtained when the reinforcement was at a depth between 0.50 and 0.75 times the foundation width. Similarly, highest ultimate bearing capacity with 2 reinforcement layers was predicted when the spacing between them was 1.0 times the width of the foundation. However, higher settlement was estimated at failure for the reinforced sloped fill than the unreinforced one. The second problem investigated was the three-dimensional (3D) analysis of Sackville embankment, a geosynthetic reinforced embankment on soft soil. Previous analyses using two-dimensional (2D) numerical modelling of Sackville embankment indicated potential 3D effects affecting the performance of this embankment. Therefore, 3D analysis incorporating geometric variations of Sackville embankment foundation soil, anisotropic model for fluid flow, mobilization of geotextile stresses in minor direction and the boundary effects (lateral directions) were taken into account in this analysis. The predicted performance of Sackville embankment were compared with the field data and the previously reported 2D analysis results in terms of vertical and horizontal displacements and excess pore pressures in the foundation soil, and geotextile stresses, strains and displacements. Better overall predictions of the Sackville embankment performance was obtained from this 3D analysis than the previous analysis reported in the literature.

Prevent-Teach-Reinforce (PTR) is a collaborative, standardized process that was developed as a way to address identified barriers to completing effective functional behavior assessments (FBAs) in public schools. Current research literature documents the effectiveness of the PTR process in decreasing problematic behaviors and increasing social skills and academic engaged time for students in kindergarten through 8th grade. In addition, PTR demonstrates high acceptability by school personnel implementing the process. While PTR has demonstrated success in schools, questions still exist regarding variables that impact the effectiveness of this process. Therefore, the current study investigated the moderating effect of prior teacher training in managing challenging behavior on the effectiveness of the Prevent-Teach-Reinforce (PTR) process. Data regarding students' social skills, behavior problems, and academic engagement were analyzed through a series of mixed factorial analyses in order to determine the effectiveness of the PTR process. Results indicate that teachers' previous preparation in dealing with behavioral problems did not moderate the effectiveness of PTR, thus indicating that the process is equally beneficial to all teachers. However, time and the implementation of PTR were found to be significant in altering trends in student outcomes. Academic engaged time was found to be significantly altered more frequently than social skills or behavior problems. Implications of the findings for using the PTR process to address problematic student behavior in schools are discussed.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2016-02-03T11:10:47Z No. of bitstreams: 1 vanderleabarretodoamaral.pdf: 1346968 bytes, checksum: 1b6fbd1a52dd7432fcc66fc8682e52e3 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2016-02-03T11:27:44Z (GMT) No. of bitstreams: 1 vanderleabarretodoamaral.pdf: 1346968 bytes, checksum: 1b6fbd1a52dd7432fcc66fc8682e52e3 (MD5) / Made available in DSpace on 2016-02-03T11:27:44Z (GMT). No. of bitstreams: 1 vanderleabarretodoamaral.pdf: 1346968 bytes, checksum: 1b6fbd1a52dd7432fcc66fc8682e52e3 (MD5) Previous issue date: 2014-08-18 / Esta pesquisa buscou avaliar os limites da implementação do Programa Reforço Escolar, criado pela Secretaria de Estado de Educação do Rio de Janeiro, em junho de 2012, nas escolas da Regional Metropolitana VII. O objetivo deste trabalho é identificar e analisar a existência de problemas na sua implementação, além de verificar as causas da falta de adesão ou da adesão parcial das escolas que não o desenvolveram. Tal Projeto visa à diminuição dos índices de reprovação em Matemática e Língua Portuguesa no ensino médio e à elevação dos resultados nas avaliações em larga escala como SAERJ, SAERJINHO, SAEB e PROVA BRASIL. Para a coleta de dados, foi aplicado um questionário a professores e gestores em duas escolas da Regional em estudo: uma que aderiu ao projeto e outra que não desenvolveu nenhuma de suas ações. Na sua análise, foram investigados os fatores extraescolares e intraescolares que dificultaram a adesão das escolas. A partir deles, apresentou-se um Plano de Ação Educacional (PAE) que propõe estratégias de divulgação do Projeto, além de reforçar a ideia de participação efetiva dos pais na vida escolar dos educandos. Ele sugere, ainda, o aprimoramento da metodologia do Projeto e levanta questões a respeito da estrutura das escolas envolvidas no processo. / This research looked for evaluate the limits of the Academic Reinforcement Program implementation, created by State Secretariat of Education of Rio de Janeiro State, in 2012 June, in Metropolitan Regional VII schools. The work aims to identify and analyze the existence of problems in program implementation, besides verify the causes of non-joining or partial joining of schools that not developed it. This project aims to reduce the taxes of failure in Mathematics and English Language in high school, and assists in elevating results in large-scale reviews as SAERJ, SAERJINHO, SAEB and PROVA BRASIL. For data collection, a questionnaire was applied to teachers and managers in two schools of studied Regional, one that joined the project and other that don’t developed any action of Scholar Reinforce. In analysis was investigated the external and internal school factors that hindered the schools joining. From these factors, presents an Educational Action Plan (PAE), proposing divulgation strategies of project, besides reinforce the idea of effective parental involvement of scholar life of students. It suggests the improvement of the project methodology and raise questions about structure of schools involved in the process.

The purpose of this experiment was to study the effects of training history on retention and re-acquisition of stimulus control of previously learned behaviors. In Phase I, two pairs of behaviors were alternately trained. Circle and touch behaviors were trained concurrently until two consecutive errorless sessions were run. Spin and down behaviors were trained together in the same manner. Probe sessions, in which all four cues were presented, were conducted each time a pair of behaviors reached this criterion. Training of one pair did not occur until the other pair had reached criterion and probe sessions were run. Despite achieving the designated criterion during training, stimulus control changed during probes. During probe sessions, errors increased under the cues that were not currently being trained. In most cases, the type of errors emitted for each cue was the same as the behavior that was trained concurrently. The number of training sessions required to reach criterion accuracy was high during the first set of sessions and decreased over the course of the experiment. In Phase II, spin and circle behaviors were trained concurrently. The number of sessions required to reach stimulus control criteria remained low, and the number of errors emitted under the spin and circle cues during probe sessions decreased. However, the number of errors increased under the touch cue. In Phase III, a reinforce-all procedure was used instead of extinction to test stimulus control. The highest frequency of errors occurred under the touch cue, but the down error was almost exclusively emitted under every cue during the last several sessions.