Dans le domaine de l'apprentissage par renforcement, la planifi ation dans les processus de décisions markoviens est une approche en ligne utilisée pour contrôler un système dont on possède un modèle génératif. Nous nous proposons d'adresser ce problème dans le cas déterministe avec espace d'action discret ou continu. Cette thèse s'attache au chapitre 2 à présenter succinctement les processus de décision markoviens puis l'apprentissage par renforcement. Nous présentons en particulier trois algorithmes centraux que sont l'itération de la valeur, l'itération de la politique et le Q-Learning. Au chapitre 3, nous expliquons l'approche de la planifi cation dans les processus de décision markoviens pour contrôler des systèmes en ligne. Ainsi, nous supposons posséder un modèle génératif d'un système à contrôler et nous l'utilisons pour décider, à chaque pas de temps du système à contrôler, de l'action à lui appliquer en vue de le faire transiter dans un état maximisant la somme future des récompenses dépréciées. Nous considérons un modèle génératif comme une boite noire, laquelle étant donnée un état et une action, nous retourne un état successeur ainsi qu'une récompense associée. L'approche optimiste est détaillée dans sa philosophie et dans son application à la résolution du dilemme exploration-exploitation au travers de di fférentes techniques présentes dans la littérature. Nous présentons di fférents algorithmes issus de la littérature et s'appliquant dans le cadre de la plani fication dans les processus de décision markoviens. Nous nous concentrons en particulier sur les algorithmes effectuant une recherche avant par construction d'un arbre des possibilités look-ahead tree en anglais. Les algorithmes sont présentés et mis en relation les uns avec les autres. L'algorithme de recherche du plus court chemin dans un graphe A est présenté en vue d'être relié à notre première contribution, l'algorithme de plani fication optimiste. Nous détaillons cette première contribution au chapitre 4. Dans un premier temps, nous présentons en détail le contexte de la planification sous contrainte de ressources computationnelles ainsi que la notion de regret. Dans un second temps, l'algorithme de plani cation uniforme est présenté et son regret est analysé pour obtenir une base comparative avec l'algorithme de plani cation optimiste. Enfi n, celui-ci est présenté et son regret est analysé. L'analyse est étendue à une classe de problèmes dé finie par la proportion de chemins -optimaux, permettant ainsi d'établir une borne supérieure sur le regret de l'algorithme de plani cation optimiste meilleure que celle de l'algorithme de plani cation uniforme dans le pire des cas. Des expérimentations sont menées pour valider la théorie et chi rer les performances de l'algorithme de plani cation optimiste par le biais de problèmes issus de la littérature comme le cart-pole, l'acrobot ou le mountain car et en comparaison à l'algorithme de plani cation uniforme, à l'algorithme UCT ainsi qu'à l'algorithme de recherche aléatoire. Nous verrons que, comme suggéré par la dé nition de la borne supérieure sur son regret, l'algorithme de plani cation optimiste est sensible au facteur de branchement ce qui nous mène à envisager le cas où l'espace d'action est continu. Ceci fait l'objet de nos deux autres contributions au chapitre 5. Notre deuxième contribution est l'algorithme de plani cation lipschitzienne reposant sur une hypothèse de régularité sur les récompenses menant à supposer que la fonction de transition et la fonction récompense du processus de décision markovien modélisant le système à contrôler sont lipschitziennes. De cette hypothèse, nous formulons une borne sur un sous-ensemble de sousespaces de l'espace d'action continu nous permettant de l'explorer par discr étisations successives. L'algorithme demande cependant la connaissance de la constante de Lipschitz associée au système à contrôler. Des expérimentations sont menées pour évaluer l'approche utilisée pour diff érentes constantes de Lipschitz sur des problèmes de la littérature comme le cart-pole, l'acrobot ou la lévitation magnétique d'une boule en acier. Les résultats montrent que l'estimation de la constante de Lipschitz est diffi cile et ne permet pas de prendre en compte le paysage local des récompenses. Notre troisième contribution est l'algorithme de plani cation séquentielle découlant d'une approche intuitive où une séquence d'instances d'un algorithme d'optimisation globale est utilisée pour construire des séquences d'actions issues de l'espace d'action continu. Des expérimentations sont menées pour évaluer cet approche intuitive pour diff érents algorithmes d'optimisation globale sur des problèmes de la littérature comme le cart-pole, le bateau ou le nageur. Les résultats obtenus sont encourageants et valident l'approche intuitive. Finalement, nous concluons en résumant les di érentes contributions et en ouvrant sur de nouvelles perspectives et extensions.

Thesis--M.A.(Ed.), University of Hong Kong. / Typewritten.

Two studies were undertaken in order to investigate the effect of different forms of error feedback and error framing in learning a complex novel task. The experimental task in both studies was a computer-based simulation of a group management situation. After each of the 12 trials, all participants received feedback about their performance on that trial. Participants receiving signal error feedback were also advised as to where they had made errors. Participants receiving diagnostic feedback were told how they could have achieved optimum performance on the previous trial. Learning, performance, strategy, exploration and depth of processing were measured during the task. Self-report measures of self-efficacy, self-set goals, satisfaction and intrinsic motivation were taken after the first six trials and again after all 12 trials were completed. In study 1, detailed diagnostic feedback was associated with better performance than feedback which simply signaled where an error had been made, or feedback that did not identify errors. Diagnostic feedback facilitated the development and use of effective problem-solving strategies and discouraged trial-and-error exploration of the problem space. In this research, exploration was found to be negatively associated with learning and performance. Learners??? self-efficacy moderated the effects of error feedback: learners with high self-efficacy showed high levels of performance regardless of the level of information that the feedback provided but for those with low self-efficacy, detailed diagnostic feedback was essential for the learning process. In the second study, positive error framing (error management) was investigated as a possible means of making signal error feedback more valuable in learning. However while positive error framing was associated with more exploration as expected, it also produced poorer strategies and worse performance than negative error framing (error avoidance instructions). Participants who used good learning strategies instead of exploration performed well despite impoverished feedback. Self-efficacy moderated the impact of error framing: positive error framing helped those with low self-efficacy, but for those with higher self-efficacy it was of more value to encourage error avoidance than error tolerance. The findings show important interactions between error framing, error feedback and learner characteristics.

Thesis (M.S.)--Rochester Institute of Technology, 1992. / Typescript. Includes bibliographical references.

Thesis (M.A.(Ed.))--University of Hong Kong. / Also available in print.

No description available.

No description available.

The purpose of this study was to investigate the impact of using novel demonstrations and hands-on activities on promoting student interest and positive perception of learning environment. Thirty-six 8th graders participated in the experimental group in which they were involved in prediction-observation-explanation of discrepant events or novel hands-on activities for 14 weeks. Their interest of learning science and perception of learning environment were compared with that of a control group, which also consisted of 36 8th graders in the same school. The results of the analysis of covariance revealed that the experimental group outperformed the control group on both instruments documenting students¡¦ level of interest in learning science and their perception of learning environment. The post-treatment interview results further consolidated the quantitative finding of the experimental group students¡¦ significant progress on their emotions of learning science.

A year long learning strategy course was designed and embedded in a Year 8 science curriculum. The Science Learning Strategy (SLS) program aimed to improve student ability to apply learning strategies to science, increase student achievement in science and to augment students' feelings of control over their science learning, so that their perceived competence was maximised. Achievement of these aims was monitored by collecting perceptions from students, parents and the teacher/researcher via a range of devices including questionnaires, work samples and interviews. The program overall was regarded as successfully achieving all three aims by 22 of the 24 students. The other two students found that only some aspects of the course were helpful, and felt they had learned little from the program. Thirty three percent of parents attributed positive changes in their daughter's study and learning strategies to participation in the SLS program (the remainder were unsure, or did not know of any changes). In relation to perception of academic performance, 38% of the parents interviewed believed that the SLS had a positive effect on their daughter's achievement in science. Several of these parents reported very positive effects on performance. The remainder were not sure or did not know if there had been any positive effects. No parents mentioned that the SLS program had caused a drop in science performance. The proportion of parents believing their daughters blamed disappointing results on factors they couldn't control dropped from 45% at the start of the year, to 22% by the end of the SLS program. After the intervention, 33% of parents reported that their daughters had come to believe that their science performance was affected by many factors, most of which they could control. / The teacher/researcher observed strong improvement in student ability to apply learning strategies to science as a result of participation in the program. Students were also observed by the teacher/researcher, to have been assisted by the intervention to realise that their science performance was governed not only by their natural ability, but also by factors such as studying behaviour and affective influences. In particular, the program appeared to the teacher/researcher to have helped students realise that they had control over their use of learning strategies, and that this control could influence their science performance. However, the teacher/researcher found that no statistically significant changes in science achievement resulted from student participation in the SLS course. The other objective of the research was to investigate the extent to which learning strategy education was valued and implemented by Western Australian science teachers. The 218 returned surveys revealed that most respondents recognised the need to teach these skills during science lessons. Seventy six percent of respondents considered the delivery of learning strategy instruction in the lower school science classroom to be as important, or more important, than teaching subject processes and content. Sixty seven percent recognised that improving study strategies can improve confidence and/or motivation. / Many teachers, however, had not been able to convert these views into consistent classroom practice. A moderate proportion of teachers reported teaching a variety of learning strategies; 74% of the respondents agreed that learning strategy instruction could improve performance in science. Only one teacher specifically mentioned incorporating the teaching of learning strategies with instruction in science process and content. As a future outcome of this project, the teacher/researcher will encourage other teachers to embed learning strategy instruction within the science curriculum, so that their students come to feel more in control of their learning and can learn more effectively.

This study evaluated a new Integrated Science Learning Environment (ISLE) that bridged the gaps between the traditionally separate classroom, field trip, and information technology milieus. The ISLE model involves a multi-faceted design to address the three basic forms of learning: acquisition of knowledge, change in emotions or feelings, and gain in physical or motor actions or performance. A holistic approach to teaching at the university level encompassed a step-wise, cumulative strategy that reinforced all scales of the Constructivist Learning Environment Survey (Personal Relevance, Uncertainty of Science, Shared Control, Critical Voice, and Student Negotiation) and minimised the potentially detrimental effects of information overload and non-linear processing. By addressing individuals and recognising limitations, the same conceptual and logistical frameworks were applied to teachers and to students uniformly in the classroom and in the field. This key factor of the ISLE program broadened all participants’ horizons and enabled them to see their role within the ‘big picture’. Thus, the common elements (knowledge) and basic components (understanding) in each realm became evident and the power of transfer for both content and concept was realised. A process approach to information technology provided a logical and meaningful mechanism for continuously scaling the program perspective from the classroom setting to the unique global environment of the World Wide Web. The final product of the ISLE program (virtual field trip) was constructed by linking the elements common to the supporting learning environments (university classroom, field trip, and information technology) at their basic levels: newness, massiveness, and appropriateness. / A combination of qualitative methods and quantitative measures provided insight into the field trip milieu and evaluation of the near- and far-term effects of exposure to constructivist pedagogy answering the general question of whether changing teachers’ learning environments might affect a change in their respective students’ learning environments. Quantitative assessment through learning environment dimensions, attitude scales, and concept map analyses was supported by qualitative data derived from reflective field journals, interviews, and observations to investigate the impact of the emergent model. Data were collected from classroom teachers and their students to assess the impact of the ISLE program in terms of promoting a constructivist classroom learning environment, teachers’ attitudes toward information technology, and teachers’ conceptual development. School teacher and student subgroups were compared in terms of the teachers’ university/field trip program experience and content background. To this end, three new versions of the Constructivist Learning Environment Survey (CLES) were shown to be valid and useful in secondary schools and graduate university courses in Texas. Data from 1079 students in 59 classes in north Texas were subjected to principal components factor analysis confirmed the factor structure, internal consistency reliability, discriminant validity, and the ability to distinguish between different classes and groups for the comparative student form (CLES-CS). Descriptive statistics supported the usefulness of the comparative teacher (CLES-CT) and adult (CLES-A) forms. Administration of these versions of the same instrument was used to characterise the learning environment of the ISLE university/field trip program, as well as the public/private school classrooms. / Further analysis and interpretation of these data suggest that the ISLE program was effective in terms of the degree of implementation of constructivist teaching approaches in the teachers’ school classrooms as assessed by teachers’ perceptions of the learning environment of their current classroom environment relative to other classes taught by them previously and students’ perceptions of the learning environment of their classroom environment relative to classes taught by other teachers in their school classrooms. Additional data suggest that the ISLE program was effective in terms of teachers’ perceptions of the university/field trip learning environment; changes in teachers’ attitudes to information technology; and teachers’ conceptual development. When an ANOVA was used to compare students’ perceptions of THIS and OTHER classes, statistically significant differences were found for some CLES scales. In particular, students whose teachers had attended the ISLE program (THIS) perceived higher levels of Personal Relevance and Uncertainty of Science in their classrooms relative to the classrooms of other teachers in the same schools (OTHER). From a practical point of view, this study documents a new model for improving learning and understanding in the field of education, specifically science education. Participation in the ISLE program provided a tangible opportunity for teachers to gain organised knowledge to make practical changes in their school classrooms. From a research point of view, this study makes a unique contribution to the field of learning environments by evaluating a comprehensive professional development program that used information technology to initiate teacher change from the central perspective of the learning environment. / Development and validation of the CLES-CS contributes to a useful range of instruments for a variety of classroom contexts within the burgeoning field of learning environments research. The real world is where theory and practice come together and science becomes relevant, making sense that leads to understanding. The conceptual and logistical frameworks of the ISLE model seamlessly merged theory and practice with science and education through effective applications of information technology to create a rich learning environment. Virtual field trips, based on the ISLE model, can enable the principles of student-centred inquiry and constructivism to be practised for the benefit of all styles and ages of lifelong learners.