Dissertations / Theses on the topic 'Computing courses'

This study investigated whether a predictive tool developed by authors in the United States (Konvalina, Stephens and Wileman) could be used with University students in Australia (in particular the Australian Capital Territory) to predict their success in first year University computing courses. It also investigated the effect of demographic and past academic factors in conjunction with, and instead of the predictive test. The study examined differences in performance between male/female students, English as a Second Language (ESL)/non-ESL students and full-time/part-time students. It also examined the effect of all the above factors on the continuing success of students in the course. While significant differences in first-time performance were found between ESL and non- ESL students, no differences were found between the other pairings. No differences were found between any of the groups in the continuing success in the course. The KSW Test, while being an indicator of first year success, was not a strong enough model to be able to be used as a predictive tool. The demographic and previous academic data from students recently at High School, in particular, the Tertiary Entrance Score, level of mathematics studied, and previous computing study, were found to be more useful as an indicator of success in fust year, explaining 53% of the variation in h a 1 unit score. In addition, 67% of the variation in continuing success in their course was also explained by the Tertiary Entrance Score, ASAT verbal and ASAT quantitative scores.

The problem addressed is the significant lack of empirical research to describe the nature and extent of technology for use in physical therapy education (PTE). The goal was to facilitate the use of instructional technologies in accredited physical therapy (PT) courses. Computing technologies offer efficient, accessible methods for delivery of education as well as instructional formats with unique advantages for the allied health sciences. In order to facilitate the use of instructional technologies in accredited PT courses the nature and extent of current technology use in PTE are described. A description of technologies used for health professional education was extracted from the literature and used to develop a valid and reliable online survey instrument. An effort was made to survey all full-time faculty in the 200 Doctor of Physical Therapy programs accredited by the Commission on Accreditation in Physical Therapy Education. A 44% response rate was received from 904 faculty representing 193 programs. The results demonstrate that the technologies used most can be characterized as those that support cognition rather than content delivery. It is also apparent that a significant number of faculty are using technology for education while their self-assessment indicates that they have insufficient knowledge and skills to do so. Also, the findings indicate that many faculty have a limited knowledge of the technologies used in the locations where their students will receive clinical education and possibly gain employment following graduation. These results are discussed in detail. Seven recommendations are offered to facilitate diffusion of technology throughout courses offered in graduate PT programs.

This longitudinal study set out to improve the retention and achievements of diverse students on computing courses in one wide access university, firstly by early identification of students at risk of poor performance and secondly by developing and implementing an intervention programme. Qualitative data were obtained using the ASSIST questionnaire, by focus group discussions and an open-ended questionnaire on students’ experiences of the transition to higher education (HE). Quantitative data on student characteristics and module results were obtained from Registry. Statistical analyses were performed using SPSS version 10. The study comprised two phases where phase one sought to enable the early detection of students at risk of poor performance by investigating the data set for patterns that may emerge between student achievement at Level 1 and entrance qualification, feeder institution, approaches to learning, conceptions of learning, course and teaching preferences and motivation. Phase one findings showed a trend of poorer performance by students who entered computing courses in HE with an AVCE entrance qualification. It was also shown that mature students scored more highly on the deep approach scale compared to their younger counterparts. Phase two investigated the data set for patterns that may emerge between student achievement at Level 2 and entrance qualification, approaches to learning, conceptions of learning and course and teaching preferences. Phase two, using action research, also sought to develop an intervention programme from the findings. This intervention programme was designed to improve aspects of information delivery to students; the personal tutor system, assessment régimes, Welcome Week, and teaching and learning. Piloting, evaluation and refinement of the intervention programme brought changes that were seen as positive by both staff and students. These changes included the Welcome Week Challenge which involved students in activities that sought to enhance students’ interactions with peers, personal tutors and the school and university facilities. These findings have shown that, for staff in wide access HE institutions, some knowledge of the previous educational experiences of their students, and the requirements of those students, are vital in providing a smooth transition to HE. A model of the characteristics of a successful student on computing courses in HE and a model for enhanced retention of diverse students on computing courses in HE were developed from the research findings. These models provide a significant contribution to current knowledge of those factors that enhance a smooth transition to HE and the characteristics of a successful student in a wide access university.

This thesis is significant in that it is one of the first evaluations of a computer classroom psychosocial learning environment and investigation of associations between learning environment factors and students' attitudes at the tertiary level in Thailand. Both quantitative and qualitative methods were used in this study. Three questionnaires were employed to provide quantitative data: the College and University Classroom Environment Inventory (CUCEI), the Computer Laboratory Environment Inventory (CLEI), and the Attitude towards Computer and Computer Courses (ACCC). The three questionnaires were administered to 905 computer science students in order to investigate their perceptions of their learning environment and associations between this and their attitudinal outcomes. Overall, the results generated from scale internal reliability analysis, mean correlations and ANOVAs suggested that the modified Thai versions of the CUCEI, CLEl, and ACCC are valid and reliable instruments for measuring students' perceptions of computing laboratory learning environments in a Thailand university. The results of an application of the CUCEI and CLEI demonstrated that students had positiveperceptions about their computer classroom learning environment. The qualitative data obtained from student interviews supported the information from questionnaires and provided more detail about the computer classrooms. Measurements of students' attitudes indicated that students enjoyed their classes and thought they were useful. Regarding associations between students' attitudes and perceptions of the computer classroom, most scales of the Thai CUCEI and CLEI, were statistically significantly positively associated with the four scales of the Thai version of the ACCC. Importantly, there were significant negative correlations between scales of the CUCEI, and CLEI with the Anxiety scale.

This study involved examining differences and similarities between the learning environment perceptions of students attending the Computer Science department and the Management department at an information technology university in Jakarta, Indonesia. In doing so, the study investigated which types of learning environments were most likely to strengthen student outcomes in computer-related courses and identify ways in which the university could enhance the teaching and learning process.The study examined whether relationships exist between students' cognitive and affective outcomes and four productivity factors: the learning environment; the quality of teacher-student interactions; students' aptitude; and students' motivation to select their chosen subject. To measure the four productivity factors, 422 students from 12 classes were asked to respond to four questionnaires that were modified to suit tertiary-level computing students: (1) the What is Happening in this Class? questionnaire (WIHIC) to measure students' perceptions of the learning environment, (2) the Questionnaire on Teacher Interaction (QTI) to measure students' perceptions of the student-teacher interaction; (3) the Test of Logical Thinking (TOLT) to measure the students' aptitude; and (4) a scale that was developed to measure students' motivation towards their course. To measure students' cognitive outcomes, information was retrieved from the university database and, to measure students' attitudes towards their computer-related subjects, four modified scales from the Test of Science Related Attitudes (TOSRA) were used.Each of the instruments was found to be valid and reliable in the Indonesian language for use at the university level in terms of factor structure, internal consistency reliability, and ability to differentiate between the perceptions of students in different classrooms. These instruments ++
provide a, means by which lecturers can monitor their classroom environments, their lecturer interaction behaviour and their students' logical thinking, motivation and attitudes. Generally, it was found that computer science students perceived the classroom environments more favourably than management students. These findings related to departmental differences at the university level provide insights into how students from different departments perceive the learning environment. Also, the study pointed to departmental differences in students' logical thinking which could influence the types of learning environment that suit students from different departments. Departmental differences in students' perceptions of the lecturer-student interpersonal behaviour suggest that lecturers should take note that the personal relationships which they build and the ways in which they treat students.

This is a study of the impacts of the incorporation, into certain points of the Computer Science degree program at the University of New Orleans, of Course Management software with an Autograding component. The software in question, developed at Carnegie Mellon University, is called “Autolab.” We begin by dissecting Autolab in order to gain an understanding of its inner workings. We can then take out understanding of its functionality and apply that to an examination of fundamental changes to courses in the time since they incorporated the software. With that, we then compare Drop, Failure, Withdrawal rate data from before and after the introduction of Autolab. With this collection of data, we can conclude, to a certain extent, that Autolab has had a negligible impact on course outcomes, but a measurable impact on course structure and pedagogy as well as improved quality of life for students and professors, alike.

This study examined the entry-level computing skills of undergraduate education majors at the University of North Carolina at Wilmington during the Spring 2005 semester. This study also compared groups based on demographic categories to investigate if certain demographics were predictors of specific skills competency. This study utilized a representative convenience sample of 186 participants. The participants were pre-tested for their ability to complete 60 computer skills in the Microsoft Office Suite using an online performance test called Skills Assessment Manager by Thomson Course Technology. The data was analyzed as a whole group performance using descriptive statistics and analyzed for analysis between demographic groups using a non-parametric statistic test (the Mann Whitney U Test). The results yielded data that informed the researcher of the skills of the participants prior to taking a mandatory computer skills course. As a result, it was found that a large portion of students were able to do many of the skills before taking the mandatory skills course. It was also found that demographics were not a reliable predictor of computer skills. This study provided data that helped to inform the faculty at UNC-W that the curriculum for the computer skills course needed to be changed based on entry skills of students to reflect the abilities of students in 2005.

This study investigated the implications that the scheduling of a block module for a computer competency course over two weeks, has had on the curriculum. The block module was presented at the start of the first semester, 2001 at the Vaal Triangle Technikon. The objectives were to: develop a flexible, outcomes-based curriculum in which the learners had to demonstrate capability and employability by integrating computer competencies in other modules; design and implement an effective learning and teaching strategy in order to ensure that learners will achieve the learning outcomes within two weeks; and to . implement continuous, integrative assessment methods in order to foster the learning of hands-on skills that can be integrated in the learners' academic programmes. The researcher followed an action research approach, in order to assess the improvement of current educational practices. A target group of first-year learners was divided into a block module-group and a semester-group. The researcher made use of data triangulation, by collecting qualitative and quantitative data comprising structured interviews and open-ended and structured questionnaires completed at different stages of the project. This research report comprises three articles. The first article reports on the high success rate of the block module-group and reveal that these learners could retain. and utilise the computer competencies that they had to utilise for completing assignments, to a larger degree than the semester-group. The second article reports that experiential learning, as an outcomes-based strategy, fosters learning in accelerated learning contexts. The third article concludes that the continuous integrative approach to assessment fosters w-operation, feedback through self-assessment and the ability of learners to apply computer competencies in new situations. It is recommended that lecturers in all the associated departments should collaboratively assign and assess tasks in progressive advanced modules. It concludes that, in order to utilise resources optimally when scheduling block modules, ail associated departments would have to be fully modularised.
Thesis (M.Ed.)--Potchefstroom University for Christian Higher Education, 2003.

La prise en considération des exigences en matière de sécurité, de vie privée et de confiance au sein des environnements pervasifs (ubiquitaires) est indispensable à la fourniture des services personnalisés aux utilisateurs. L’objectif de cette thèse est de disposer d’une architecture souple et évolutive intégrant l’authentification des utilisateurs, la préservation de leur vie privée et la gestion de la confiance en vue d’optimiser la stratégie de contrôles d’accès aux services personnalisés. La première contribution porte sur la proposition d’un protocole d’authentification mutuelle construit à partir de schémas cryptographiques robustes d’établissement de clés basés sur les courbes elliptiques (MaptoPoint/Curve algorithm, Weil Pairing) et d’un modèle dynamique basé sur les attributs issus des données contextuelles. La seconde contribution porte sur la conception d’une nouvelle architecture bâti sur un modèle basé sur les attributs et organisée autour de 3 couches : la couche de contrôle de le vie privée qui assure la protection de la vie privée des utilisateurs en contrôlant leurs données personnelles, la couche d’accès associant les processus d’authentification et de contrôles d’accès en intégrant des mécanismes dédiés à la gestion des paramètres de confiance et la couche de service pour la gestion des accès aux services selon le profil de l’utilisateur et de son environnement. La troisième contribution porte sur le développement et la mise en œuvre d’un prototype au sein de la plateforme dédiée à la fourniture de services du laboratoire Handicom de Telecom SudParis
To provide personalized services to users in pervasive environments, we should consider both user's privacy, trust and security requirements. Traditional authentication and access control mechanisms are not able to adapt their security policies to a changing context while insuring privacy and trust issues. This thesis introduces a new global vision for the protection of pervasive environments, based on context-aware principle. The aim of this thesis is to get a flexible and scalable framework including user authentication, user privacy preserving and trust management in order to optimize the access control strategy to personalized services. The first contribution include the proposal of a mutual authentication protocol supported by both robust key establishment schemes based on elliptic curves (MaptoPoint/Curve algorithm, Weil Pairing) and a dynamic model based on attributes issued from contextual data. The second contribution include the design of a new architecture built on an attribute based model and organized over 3 layers: the privacy control layer which insure the protection of the user private life by controlling their personal data, the access layer which associate authentication and access control processes while providing mechanisms dedicated to trust parameters management , and finally the service layer for service access management according to the user profile and his environment. The third contribution the implementation and the deployment of a prototype within the service delivery platform in Handicom lab of Telecom & Management SudParis

L'objectif de notre thèse est de contribuer à la conception de situations informatiques diffuses efficaces énergétiquement et appropriables par les habitants au quotidien. Pour cela, à partir d'une approche centrée sur l'activité, nous caractérisons les contextes d'activité quotidienne, ce qui permet d'élaborer nos principes de conception et des scénarios. Cette recherche est menée en référence au cadre du " cours d'action " (Theureau, 2004) ainsi qu'à une approche critique et située (e.g., Barad, 2007; Suchman, 2007) de l'agentivité physique sous-tendant l'action collective, via la notion de " cours d'agencement ". Les spécificités de l'informatique diffuse (e.g., en réseau disséminé dans l'environnement) et des situations domestiques investiguées (e.g., plusieurs pièces et étages) nous amènent à créer une méthodologie pour l'analyse formelle du collectif et les scénarios : le format " multi-portées ". Sur ces bases, nous caractérisons les contextes d'activité de plusieurs familles vivant dans des maisons à forte consommation énergétique. Nos résultats mettent en évidence les préoccupations des habitants, les jeux de suspension/actualisation à travers le temps, ou de préoccupations multiples à l'instant " t " et les anticipations qui les structurent. Nous caractérisons également des contextes d'articulations collectives, montrant que les membres de la famille sont engagés dans des préoccupations à la fois similaires et différentes, oscillant entre convergence et divergence à différentes échelles de temps. Enfin, nous comparons 85 agencements du collectif dans l'habitat et les types d' " (im)possibilités " perceptives et pratiques qui en émergent entre humains, objets et espaces physiques. Ces résultats permettent d'alimenter notre modèle de conception de situations informatiques diffuses appropriables et efficaces énergétiquement. Après avoir présenté les orientations et principes de conception généraux, nous construisons des scénarios collectifs d'" interactions incidentes/intentionnelles " (Dix, 2002) entre humains et réseaux, puis nous définissons comment le futur système devra fonctionner afin d'être appropriable et efficace énergétiquement. Ce travail montre qu'il est possible et nécessaire de concevoir l'efficience énergétique des réseaux socio-techniques tout en donnant des possibilités d'action et d'appropriation à l'humain.

Šiame magistro darbe mes analizuojame nuotolinių studijų kurso bendrai ir mokymo medžiagos atskiromis kurso dalimis pertvarkymą. Reinžinerijos koncepcinis modelis yra įvairiai interpretuojamas ir yra taikomas programinės įrangos rengimo, ar vadybos moksluose. Tai yra daroma, kad sistemas būtų galima geriau panaudoti. Jos analizuojamos ir pertvarkomos tolimesniam naudojimui. Mes sujungiame minėtų sričių reižinerijos koncepto reikšmes ir naudojame tai kaip metodinį pagrindą nuotolinių studijų srityje. Mes analizuojame nuotolinių studijų kurso struktūrą. Tada pristatome kursus paruoštus ir naudojamus užsienio universitetuose, susijusius su ‘Programavimu grafinėje terpėje‘, trigubo pastovumo principu, Bloom taksonomija ir jos pritaikymu kompiuterijos mokslų studijavimui, informatikos mokymo programa 2001 (angl. computing curriculla 2001). Toliau, mes siūlome konceptualią nuotolinių studijų pertvarkos struktūrą iš dėstytojo perspektyvų ir pristatome atvejų analizę, kurioje pora temų yra pertvarkytos, atsižvelgiant į trigubo pastovumo principą ir reikalavimus kompiuterių mokslų studentams.
In this master thesis we analyse a problem of reengineering of a distance study system, in general, and the learning material of a separate course, in particular. Reengineering concept with its different interpretations is used in software engineering and management sciences. It deals with making systems better maintainable, examination and reconstitution of the system for further reimplementation. We combine the meaning of reengineering concept in both mentioned areas and employ it as methodological background in distance study area. We analyse the structure of the distance study course. Then we introduce the courses, prepared and delivered in foreign universities and related to ‘Programming in GUI’ course, triple consistency principle, Bloom taxonomy and its applicability to computer science studies, Computing Curricula 2001. Further, we propose a conceptual distance study course reengineering framework from the lecturer’s perspective and present a case study, in which two topics were reengineered, considering triple consistency principle and requirements for computer science students.

Les travaux présentés s'inscrivent dans le cadre des systèmes mobiles et distribués, et s'intéressent tout particulièrement aux perspectives offertes par les réseaux locaux sans fil. A l'opposé de la complexité de déploiement d'une infrastructure cellulaire étendue, les interactions sans fil courte portée peuvent être utilisées de manière très simple, sans infrastructure. Ainsi, elles permettent à des calculateurs proches d'échanger automatiquement des informations. Nous proposons des supports système prenant en compte la volatilité des communications sans fil, et permettant de développer des applications tirant spontanément parti de la proximité physique des noeuds mobiles. Ces travaux sont ensuite étendus dans le cadre d'autres familles de réseaux sans fil. Ainsi, nous nous intéressons aux réseaux à couverture discontinue. La technologie support est la même que celle de notre première étude. Simplement, les communications entre les noeuds mobiles ne sont plus directes, mais passent par une borne fixe. Cette borne définit une bulle radio de taille limitée. C'est l'interconnexion de ces bulles, sans souci de continuité de la couverture radio, qui permet d'envisager un réseau étendu et simple à déployer. Dans ce cadre, les mécanismes système étudiés permettent de masquer l'intermittence de la connectivité, et autorisent le support d'applications exploitant les flux montants et descendants dans le réseau. Enfin, dans une dernière partie, nous abordons le problème du couplage système de deux architectures sans fil hétérogènes. Un tel couplage présente des objectifs comparables à ceux des réseaux à couverture discontinue : offrir des nouveaux services sur une couverture large, à des densités importantes d'utilisateurs mobiles. Ces travaux débouchent sur la définition de mécanismes permettant de coupler au sein d'un même service des propriétés fonctionnelles des deux infrastructures sans fil.

Les protocoles de cryptographie asymétrique nécessitent des calculs arithmétiques dans différentes structures mathématiques. Un grand nombre de ces protocoles requièrent en particulier des calculs dans des structures finies, rendant indispensable une arithmétique modulaire efficace. Ces opérations modulaires sont composées d'opérations multiprécision entre opérandes de tailles suffisamment grandes pour garantir le niveau de sécurité requis (de plusieurs centaines à plusieurs milliers de bits). Enfin, certains protocoles nécessitent des opérations arithmétiques dans le groupe des points d'une courbe elliptique, opérations elles-mêmes composées d'opérations dans le corps de définition de la courbe. Les tailles de clés utilisées par les protocoles rendent ainsi les opérations arithmétiques coûteuses en temps de calcul. Par ailleurs, les architectures grand public actuelles embarquent plusieurs unités de calcul, réparties sur les processeurs et éventuellement sur les cartes graphiques. Ces ressources sont aujourd'hui facilement exploitables grâce à des interfaces de programmation parallèle comme OpenMP ou CUDA. Cette thèse s'articule autour de la définition d'opérateurs arithmétiques parallèles permettant de tirer parti de l'ensemble des ressources de calcul, en particulier sur des architectures multicœur à mémoire partagée. La parallélisation au niveau arithmétique le plus bas permet des gains modérés en termes temps de calcul, car les tailles des opérandes ne sont pas suffisamment importantes pour que l'intensité arithmétique des calculs masque les latences dues au parallélisme. Nous proposons donc des algorithmes permettant une parallélisation aux niveaux arithmétiques supérieurs : algorithmes parallèles pour la multiplication modulaire et pour la multiplication scalaire sur les courbes elliptiques. Pour la multiplication modulaire, nous étudions en particulier plusieurs ordonnancements des calculs au niveau de l'arithmétique modulaire et proposons également une parallélisation à deux niveaux : modulaire et multiprécision. Ce parallélisme à plus gros grain permet en pratique des gains plus conséquents en temps de calcul. Nous proposons également une parallélisation sur processeur graphique d'opérations modulaires et d'opérations dans le groupe des points d'une courbe elliptique. Enfin, nous présentons une méthode pour optimiser la multiplication scalaire sur les courbes elliptiques pour de petits scalaires.

Dans cette thèse, nous développons des algorithmes à haute performance pour certains calculs impliquant des tenseurs denses et des matrices éparses. Nous abordons les opérations du noyau qui sont utiles pour les tâches d'apprentissage de la machine, telles que l'inférence avec les réseaux neuronaux profonds. Nous développons des structures de données et des techniques pour réduire l'utilisation de la mémoire, pour améliorer la localisation des données et donc pour améliorer la réutilisation du cache des opérations du noyau. Nous concevons des algorithmes parallèles à mémoire séquentielle et à mémoire partagée.Dans la première partie de la thèse, nous nous concentrons sur les noyaux tenseurs denses. Les noyaux tenseurs comprennent la multiplication tenseur-vecteur (TVM), la multiplication tenseur-matrice (TMM) et la multiplication tenseur-tendeur (TTM). Parmi ceux-ci, la MVT est la plus liée à la largeur de bande et constitue un élément de base pour de nombreux algorithmes. Nous proposons une nouvelle structure de données qui stocke le tenseur sous forme de blocs, qui sont ordonnés en utilisant la courbe de remplissage de l'espace connue sous le nom de courbe de Morton (ou courbe en Z). L'idée clé consiste à diviser le tenseur en blocs suffisamment petits pour tenir dans le cache et à les stocker selon l'ordre de Morton, tout en conservant un ordre simple et multidimensionnel sur les éléments individuels qui les composent. Ainsi, des routines BLAS haute performance peuvent être utilisées comme micro-noyaux pour chaque bloc. Les résultats démontrent non seulement que l'approche proposée est plus performante que les variantes de pointe jusqu'à 18%, mais aussi que l'approche proposée induit 71% de moins d'écart-type d'échantillon pour le MVT dans les différents modes possibles. Enfin, nous étudions des algorithmes de mémoire partagée parallèles pour la MVT qui utilisent la structure de données proposée. Nos résultats sur un maximum de 8 systèmes de prises montrent une performance presque maximale pour l'algorithme proposé pour les tenseurs à 2, 3, 4 et 5 dimensions.Dans la deuxième partie de la thèse, nous explorons les calculs épars dans les réseaux de neurones en nous concentrant sur le problème d'inférence profonde épars à haute performance. L'inférence sparse DNN est la tâche d'utiliser les réseaux sparse DNN pour classifier un lot d'éléments de données formant, dans notre cas, une matrice de caractéristiques sparse. La performance de l'inférence clairsemée dépend de la parallélisation efficace de la matrice clairsemée - la multiplication matricielle clairsemée (SpGEMM) répétée pour chaque couche dans la fonction d'inférence. Nous introduisons ensuite l'inférence modèle-parallèle, qui utilise un partitionnement bidimensionnel des matrices de poids obtenues à l'aide du logiciel de partitionnement des hypergraphes. Enfin, nous introduisons les algorithmes de tuilage modèle-parallèle et de tuilage hybride, qui augmentent la réutilisation du cache entre les couches, et utilisent un module de synchronisation faible pour cacher le déséquilibre de charge et les coûts de synchronisation. Nous évaluons nos techniques sur les données du grand réseau du IEEE HPEC 2019 Graph Challenge sur les systèmes à mémoire partagée et nous rapportons jusqu'à 2x l'accélération par rapport à la ligne de base
In this thesis, we develop high performance algorithms for certain computations involving dense tensors and sparse matrices. We address kernel operations that are useful for machine learning tasks, such as inference with deep neural networks (DNNs). We develop data structures and techniques to reduce memory use, to improve data locality and hence to improve cache reuse of the kernel operations. We design both sequential and shared-memory parallel algorithms. In the first part of the thesis we focus on dense tensors kernels. Tensor kernels include the tensor--vector multiplication (TVM), tensor--matrix multiplication (TMM), and tensor--tensor multiplication (TTM). Among these, TVM is the most bandwidth-bound and constitutes a building block for many algorithms. We focus on this operation and develop a data structure and sequential and parallel algorithms for it. We propose a novel data structure which stores the tensor as blocks, which are ordered using the space-filling curve known as the Morton curve (or Z-curve). The key idea consists of dividing the tensor into blocks small enough to fit cache, and storing them according to the Morton order, while keeping a simple, multi-dimensional order on the individual elements within them. Thus, high performance BLAS routines can be used as microkernels for each block. We evaluate our techniques on a set of experiments. The results not only demonstrate superior performance of the proposed approach over the state-of-the-art variants by up to 18%, but also show that the proposed approach induces 71% less sample standard deviation for the TVM across the d possible modes. Finally, we show that our data structure naturally expands to other tensor kernels by demonstrating that it yields up to 38% higher performance for the higher-order power method. Finally, we investigate shared-memory parallel TVM algorithms which use the proposed data structure. Several alternative parallel algorithms were characterized theoretically and implemented using OpenMP to compare them experimentally. Our results on up to 8 socket systems show near peak performance for the proposed algorithm for 2, 3, 4, and 5-dimensional tensors. In the second part of the thesis, we explore the sparse computations in neural networks focusing on the high-performance sparse deep inference problem. The sparse DNN inference is the task of using sparse DNN networks to classify a batch of data elements forming, in our case, a sparse feature matrix. The performance of sparse inference hinges on efficient parallelization of the sparse matrix--sparse matrix multiplication (SpGEMM) repeated for each layer in the inference function. We first characterize efficient sequential SpGEMM algorithms for our use case. We then introduce the model-parallel inference, which uses a two-dimensional partitioning of the weight matrices obtained using the hypergraph partitioning software. The model-parallel variant uses barriers to synchronize at layers. Finally, we introduce tiling model-parallel and tiling hybrid algorithms, which increase cache reuse between the layers, and use a weak synchronization module to hide load imbalance and synchronization costs. We evaluate our techniques on the large network data from the IEEE HPEC 2019 Graph Challenge on shared-memory systems and report up to 2x times speed-up versus the baseline

Les travaux récents en fondements de la théorie quantique (des champs) et en information quantique relativiste tentent de mieux comprendre les effets des contraintes de causalité imposées aux opérations physiques sur la structure des corrélations quantiques. Le premier chapitre de cette thèse est consacré à l'étude des implications conceptuelles de la non-localité quantique, notion qui englobe celle d'intrication dans un sens précis. Nous détaillons comment les récentes approches informationnelles tentent de saisir la structure des corrélations non-locales, ainsi que les questions que ces dernières soulèvent concernant la capacité d'un observateur localisé à isoler un système de son environnement. Le second chapitre détaille les effets de l'invariance de Poincaré sur la détection et la quantification de l'intrication. Cette invariance impose que tous les systèmes soient modélisés en dernière instance dans le cadre de la théorie des champs, ce qui implique qu'aucun système à énergie finie ne puisse être localisé, ainsi que la divergence de toute mesure d'intrication pour des observateurs localisés. Nous fournissons une solution à ces deux problèmes en démontrant l'équivalence générique qui existe entre une résolution spatiale finie des appareils de mesure et l'exclusion des degrés de liberté de haute énergie de la définition du système observé. Cette équivalence permet une interprétation épistémique du formalisme quantique standard décrivant les systèmes localisés non-relativistes et leurs corrélations, clarifiant ainsi l'origine des mesures finies d'intrication pour de tels systèmes. Le dernier chapitre explore un cadre théorique récemment introduit qui prédit l'existence de corrélations quantiques sans ordre causal défini. Procédant par analogie avec le cas des corrélations non-locales, nous présentons quelques principes informationnels contraignant la structure de ces corrélations dans le but de mieux en comprendre l'origine physique
Recent works in foundations of quantum (field) theory and relativistic quantum information try to better grasp the interplay between the structure of quantum correlations and the constraints imposed by causality on physical operations. Chapter 1 is dedicated to the study of the conceptual implications of quantum nonlocality, a concept that subsumes that of entanglement in a certain way. We detail the recent information-theoretic approaches to understanding the structure of nonlocal correlations, and the issues the latter raise concerning the ability of local observers to isolate a system from its environment. Chapter 2 reviews in what sense imposing Poincaré invariance affects entanglement detection and quantification procedures. This invariance ultimately forces a description of all quantum systems within the framework of quantum field theory, which leads to the impossibility of localized finite-energy states and to the divergence of all entanglement measures for local observers. We provide a solution to these two problems by showing that there exists a generic equivalence between a finite spatial resolution of the measurement apparatus and the exclusion of high-energy degrees of freedom from the definition of the observed system. This equivalence allows for an epistemic interpretation of the standard quantum formalism describing nonrelativistic localized systems and their correlations, hence a clarification of the origin of the finite measures of entanglement between such systems. Chapter 3 presents a recent theoretical framework that predicts the existence of correlations with indefinite causal order. In analogy to the information-theoretic approaches to nonlocal correlations, we introduce some principles that constrain the structure of such correlations, which is a first step toward a clear understanding of their physical origin

博士
國立高雄師範大學
科學教育研究所
101
The aim of this study is to explore the transformation of cloud computing concept to cooperative learning and problem solving strategies with qualities similar to cloud computing. We purpose to understand the information course learning outcomes and cloud computing cooperative learning styles of college students under cloud computing cooperative learning scenario, as well as discuss the relationship between their cloud computing cooperative learning styles and learning outcomes. This is a case study based on research subjects selected with convenience sampling from the researcher’s classes. In the pilot study phase, the case is a freshman class (equivalent to juniors in universities) with 31 students from a two-year technology institute in southern Taiwan. A sophomore class (seniors in universities) with 28 students is the research subject the following semester. In the formal research, the case is a sophomore class (sophomores in universities) with 36 students from the same school’s junior college department. The instruments used for data collection are IP addressing test, scale for cloud computing cooperative learning style, scale for cloud computing cooperative learning behavior and attitude, semi-interviews of IP addressing, cloud platform, and discussion record of IP addressing. Descriptive statistics, paired-sample t-test, correlation analysis and qualitative data analysis are carried out next. Research results and finds are as follows: 1. the post-test scores of IP addressing learning are higher than the pre-test, indicating cloud computing cooperative learning is effective. The results of paired t-test on the pre- and post-tests show a significance of .000, among which the significances of factual choice (multiple-choice) and matrix of procedural concept are both .000, indicating cloud computing cooperative learning is significant in the factual knowledge and conceptual knowledge of information courses. 2. Cloud computing cooperative learning includes four styles: interactive reflection – cloud solving; knowledge sharing – class solving; information cognition – class ask; and data acquisition – cloud ask. Cloud computing cooperative learning consists of cycle interactive learning forms. This study uses four learning forms: (1) Data acquisition: ask first and then obtain information or data rapidly through multiple approaches. (2) Information cognition: internalized the acquired Information into knowledge. (3) Knowledge sharing: after learner internalizes knowledge, he/she can help others in problem solving, thus share knowledge. (4) Interactive reflection: learner reflects from interactions in cloud computing cooperative learning and creates new cognitions. 3. The correlation between cloud computing cooperative learning styles and learning outcomes is not significant when distinguished with quadrants, but the correlation between the three categories, nine forms and learning outcomes is significant when distinguished by radar chart. Furthermore, the learning outcomes of students implementing the interactive reflection style are better than the students performing other styles.

The improvement of throughput rates is a crucial factor at higher education institutions; hence, university departments focus on improving pass rates per module. This study investigated how lecturers in the School of Computing (SoC) at the University of South Africa, use technology for improving throughput rates in an Open Distance Learning (ODL) context. The study sought answers to the main research question on how lecturers in the SoC use technology for improving throughput rates in an ODL institution. A mixed research methods approach was used, where quantitative data was extracted from the university systems and integrated with the qualitative data collected from interviews. Thirteen lecturers for the thirty modules under investigation were interviewed. A thematic analysis was used on the qualitative data, and quantitative data was analysed using rankings and correlation coefficients, leading to the interpretation that the use of myUnisa technology assisted to improve throughput on the modules.
Mathematics Education
M. Sc. (Computing Education)

碩士
國立臺灣師範大學
資訊教育研究所
90
Course Scheduling Problem is an NP-Complete Problem, however, it is also a necessary administration task for every school in every semester. The constraints of a Course Scheduling Problem include complicated parameters such as courses, teachers, classrooms, classes and facilities in a school. It is very difficult to develop an efficient computer system to solve this kind of problem. This paper proposes a modified genetic algorithm to solve the Course Scheduling Problem, which can adapt these complicated parameters very easily and solve the problem efficiently. In order to improve the execution performance of the system, we also introduce genetic agent computing concept into our computing mechanism, which can provide concurrency computation through distributed system. We propose two genetic agent computing models: Message Queue and Collection. We find that the multi-thread and multi-process versions of genetic agent computing indeed can improve the execution performance of our system.

M.Ed.
Computers are an integral part of life, and have become integral to everyday functions of work, travel, health, entertainment and study. It is crucial that this technology be utilised to maximum effect. In schools, learners are increasingly required to enhance their computer skills, and at tertiary level students explore ways they can be used to develop and enhance capacity, functions and responsibilities. This study examines ways in which computer technology can be used to teach, so as to benefit all role-players in the learning process, including lecturers, students, the learning fraternity and industry, contributing to a more competent and capacitated workforce. In particular, it looks at an electronic assessment tool implemented by the University of Johannesburg (UJ), intended to enhance and upgrade learning in courses where computers are used. However, with the initial use of system, certain shortcomings were identified which affected its performance. As computer-related testing was considered important in contributing to assessment, it was critical to identify any problems that were inhibiting its optimal performance. To this end, the researcher studied the subdepartment End-User Computing (EUC), located within the department of Business Information Technology (BIT) at UJ. A qualitative phase consisted of two identified focus groups made up of specific individuals from BIT, comprising lecturers engaged in the daily use and activation of the electronic assessment system and supported by information system technical support personnel. A second phase, quantitative in nature, used responses to structured questionnaires delivered to students who had used the assessment tool. The University’s Statcon unit analysed results, and identified problems. Lecturers pointed to the language used in questions and system challenges to teaching and learning, while students highlighted a need for preparing them for the process, particularly calling for a tutorial with class presentation prior to assessment.

碩士
國立臺灣師範大學
科技應用與人力資源發展學系
100
The Cloud computing was reconstructed by the traditional Client-Server model, however, Cloud computing is a highly flexible computing architecture nowadays. In the past, when you need to create a learning system, building the website and other related hardware are necessary pre-tasks, which including to maintenance equipment and many of functionality servers to keeping appropriate manpower, costs of money and time. Now the Cloud computing tech has been available as high availability and flexible service of learning system, you may increase study experiences and students’ creativity. This system will be easy to upload and manage their multimedia files. And the OpenCourseWare of learning system private cloud which also supporst the open formats files; reaching the same experience between standard computer and mobile devices, to expand effectiveness of the OpenCoursesWare for education. This thesis will be implemented an private cloud service model orientation, build a CMS platform for Chinese language teachers for easy upload and management of multimedia teaching materials, the cloud CMS system also provides an open format packaged conversion function to achieve as an open format for mobile learning and portable those multimedia materials. The system may support for flexible expansion to thousands of people while on-line Chinese language learning which is related to the user by the cloud CMS system that increase the satisfaction for Chinese language learning.

Innovations in technology are changing not only everyday life for many individuals around the world but are also influencing the expansion of online learning opportunities at an accelerated rate (Collins & Halverson, 2018; Mah, 2016). Online learning platforms allow for scalability, flexibility, greater global access, and innovative and new ways to deliver education (Goodman, Melkers, & Pallais, 2019; Kizilcec et al., 2019). Enrollments in online learning programs and opportunities have seen significant growth in recent years (Seaman, Allen, & Seaman, 2018; U.S. Department of Education, 2018) with continued and steady growth expected into the future. The ubiquity and newness of new online learning formats present a challenge in linking research and practice. Through three separate academic papers, the following dissertation discusses and considers key questions and topics with regards to the use of digital badges and Massive Open Online Courses (MOOCs), two types of emerging online innovations and platforms, and aspects of their efficacy. The three papers respectively 1) identify and discuss the theoretical and empirical foundations digital badges use in specific learners groups by reviewing current literature; 2) highlight the application of a use case in which digital badges have been implemented as a means to offer training; and 3) explore the perceptions of MOOC instructors toward quality learning in their courses in a case study. Conclusions are drawn and solutions as well as potential future directions for research and practice of discussed.

碩士
國立臺南大學
數位學習科技學系碩士班
99
Affective tutoring system (ATS) is uses the factor of affective recognition with intelligent tutoring system (ITS). The aim of this study is to improve learning interests by recognizing the emotion states of students during learning and giving adequate feedback. This study consists of three research stages: (1) Design both the emotion recognition system and the tutoring strategy module. (2) Design the digital arts learning content module, the emotion feedback mechanism via the HCI design of interactive agent dolls.(3) Integrate and evaluate the whole system by two-stage evaluation. We hope that the learners’ motivations and interests could be enhanced via affective interaction design, and hence their learning performance could be improved. The system evaluation processes of this study adopts: (1) Prototype evaluation: The method of evaluation combines a system usability scale (SUS) questionnaire and heuristic evaluation by experts. (2) Triangulation evaluation: This method of evaluation uses both qualitative and quantitative research that includes observation, questionnaires and interview. By the above method, the following points are investigated: (1) Is the usability of ATS good or bad. (2) How about the satisfaction of ATS users. (3) Whether the interactivity of ATS is attractive to users. (4) Whether ATS increased the motivation of learning in digital art or not. (5) Whether ATS has different impact on self-consciousness learning achievement for users from different domains. According the result of evaluation, we can find the usability of system was rated high by users with high satisfaction ratings. Furthermore, ATS is not only attractive to users, but also increases learning motivation and self-consciousness learning achievement.