Algorithms that learn sets of rules describing a concept from its examples have been widely studied in machine learning and have been applied to problems in medicine, molecular biology, planning and linguistics. Many of these algorithms used a separate-and-conquer strategy, repeatedly searching for rules that explain different parts of the example set. When examples are scarce, however, it is difficult for these algorithms to evaluate the relative quality of two or more rules which fit the examples equally well. This dissertation proposes, implements and examines a general technique for modifying rule evaluation in order to improve learning performance in these situations. This approach, called Description-based Evaluation Function Transfer (DEFT), adjusts the way rules are evaluated on a target concept by taking into account the performance of similar rules on a related support task that is supplied by a domain expert. Central to this approach is a novel theory of task similarity that is defined in terms of syntactic properties of rules, called descriptions, which define what it means for rules to be similar. Each description is associated with a prior distribution over classification probabilities derived from the support examples and a rule's evaluation on a target task is combined with the relevant prior using Bayes' rule. Given some natural conditions regarding the similarity of the target and support task, it is shown that modifying rule evaluation in this way is guaranteed to improve estimates of the true classification probabilities. Algorithms to efficiently implement Deft are described, analysed and used to measure the effect these improvements have on the quality of induced theories. Empirical studies of this implementation were carried out on two artificial and two real-world domains. The results show that the inductive transfer of evaluation bias based on rule similarity is an effective and practical way to improve learning when training examples are limited.

Notre thèse se situe dans le domaine des méthodes formelles appliquées aux systèmes réactifs. Nous modélisons et traitons ces systèmes, en continuelle interaction avec leur environnement, grâce au langage<br />synchrone Lustre.<br /><br />D'abord, sur la base d'un travail précurseur, nous établissons pour Lustre une méthode de preuve inductive des propriétés de sûreté. Cette méthode est optimisée, afin de prendre en compte au mieux la dynamique des systèmes. Elle est implémentée dans un outil de preuve, Gloups.<br /><br />Ensuite, suivant le modèle de la méthode B, nous définissons un calcul de raffinement pour Lustre. Ce calcul est à la fois adapté à Lustre et exprimé en ce langage. Les obligations de preuve qui assurent la<br />correction du raffinement peuvent être traitées par Gloups. Pour faciliter le développement, un autre outil, Flush, génère automatiquement les obligations pour Gloups.<br /><br />Ainsi, nous utilisons Lustre à la fois comme langage de programmation et comme cadre formel d'un développement maîtrisé. L'intérêt de ce<br />procédé réside dans la simplicité du langage et dans son adaptation aux systèmes réactifs : en ce domaine, notre méthode de raffinement est suffisamment expressive, sans être inutilement compliquée. Des exemples viennent démontrer l'intérêt de la méthode.

<p>Pulsed power generators are a key component in compact systems for generation of high-power microwaves (HPM). HPM generation by virtual cathode devices such as Vircators put high demands on the source. The rise time and the pulse length of the source voltage are two key issues in the generation of HPM radiation. This thesis describes the construction and tests of several inductive high power pulse generators. The pulse generators were designed with the intent to deliver a pulse with fast rise time and rectangular voltage. Therefore, the pulse generators are all based on discharges of transmission lines. The transmission lines used in the pulse generators are modern high voltage cables normally used in power transmission. All tested pulse generators have step up transformers as intermediate magnetic energy storage. Further, the pulse generators have contributed to the knowledge of compact designs. Closing switches is another important device in pulse generator architectures. In general, high power switching involving large currents and high voltages requires the use of robust switches. The switches used in the pulse generators of this thesis are of gas and liquid type.</p>

Active magnetic bearings (AMBs) have become a key technology in industrial applications with a continued drive for cost reduction and an increase in reliability. AMBs require position feedback to suspend the rotor. Conventional contactless position sensors are used to measure the rotor's position. The major disadvantages of conventional position sensors are their cost and that the sensors are viewed as a weak point in an AMB system. A self-sensing sensor is a type of sensor which is cost effective, reduces sensor wire-length and increases reliability, thus ideal for the industry. This type of sensor relies on the current and voltage signals of the AMB's to obtain the rotor position. Due to the rapid and advanced development of digital electronics, it has become more powerful and cheaper, thus self-sensing in mass production will be cost effective. Different self-sensing approaches were developed in the past and can be divided into two main categories: state estimation and amplitude modulation approaches. In this research the focus will be on the amplitude modulation approach. Amplitude modulation makes use of two signals, namely the modulation signal and the carrier signal. In a self-sensing AMB system the carrier can be a high frequency component injected into the system or the switching ripple of the switch mode power amplifier can be used. The modulation signal is the change in rotor position which results in changing inductances. The actuator material introduces nonlinear effects on the estimated position. Due to these nonlinear effects, it is rather difficult to obtain the rotor position. The first industrial application of a self-sensing turbomolecular pump system was implemented in 2005 by S2M. The aim of this thesis is to evaluate existing self-sensing schemes, devise improvements and investigate possible new schemes. Four different demodulation methods and two new self-sensing schemes are evaluated. An AMB transient simulation model which includes saturation, hysteresis, eddy currents and cross-coupling is used to evaluate the schemes in simulation. The self-sensing schemes are implemented in hardware and evaluated on a 7 A rms 500 N AMB. A comparative study was done on the different self-sensing schemes. From the comparative study it was determined that the gain- and phase effects have a direct effect on the sensitivity of the system. It was also proved that self-sensing can be implemented on a coupled AMB with a sensitivity of 10.3 dB. / Thesis (Ph.D. (Electrical and Electronic Engineering))--North-West University, Potchefstroom Campus, 2009.

Social violence is a serious problem in Latin America, an assertion that is thoroughly supported by statistics that identify Latin America as one of the most violent regions in the world (?Searching for Solutions in Latin America and the Caribbean? 20). These violent statistics explored in the very first chapter set the stage for this proposal by establishing its dire purpose. Formative historical legacies are central to social development, and they can provide the root causes and resentments that spur such social violence. There are a few that are common to the area of Latin America as a whole, such as the colonial experience and the slave trade, and these are examined alongside legacies of key leaders and educators that have shaped the face of the region and produced both pillars of hope and some of the greatest obstacles to social change. Understanding the root causes of violence and the formidable obstacles against effecting social change are key elements in designing and instituting a solution to this problem. Within this solution, education is a universally valid channel that, if given the necessary resources and personnel, could affect the lives of the majority of citizens. Combining the ideas of Martin L. Hoffman on the effect of inductive discipline encounters on children through character education, education becomes a powerful tool for creating not only more intelligent and autonomous citizens, but more empathic ones more attuned to an ?ethic of care.? While such a proposal certainly offers no guarantees and its successful institution would need to co-exist with other much needed social, political and economic reforms, it is presented as an innovative and experimental solution to a pressing problem taking a toll on the social and economic capital of societies in the region; a toll that is not easily ignored, particularly those for whom violence is an every day reality.

In this thesis the nonlinear effects and heating are studied in inductively coupled plasma (ICP) in a regime of anomalous skin effect (nonlocal regime). In this regime the thermal motion of plasma electrons plays an important role, significantly influencing the processes associated with the penetration of electromagnetic field into plasma, such as the ponderomotive effect and heating of plasma by the field. We have developed a linear kinetic theory that describes the electron dynamics in ICP taking into account the electron thermal motion and collisions of electrons. This theory yields relatively simple expressions for the electron current in plasma, the ponderomotive force, and plasma heating. It describes correctly the thermal reduction of ponderomotive force in the nonlocal regime, which has been previously observed experimentally. It also describes the collisionless heating of plasma due to resonant interaction between the electromagnetic wave and plasma electrons. There is a good overall agreement of the results of our theory with the experimental data on ponderomotive force and plasma heating. Using our theory, we predicted a new effect of reduction of plasma heating compared to the purely collisional value, occurring at low frequencies. This effect has not been previously reported. The nonlinear effects of the electromagnetic field on the electron distribution function and on plasma heating, that are not accounted for in the linear kinetic theory, have been studied using a quasilinear kinetic theory, also developed in this thesis. Within the quasilinear approximation we have formulated the system of equations describing the slow response of plasma electrons to the fast oscillating electromagnetic field. As an example, these equations have been solved in the simplest case of cold plasma with collisions, and the nonlinear perturbation of the electron distribution function and its effect on the plasma heating have been found. It has been shown that the nonlinear modification of plasma heating occurs mainly due to the nonlinear effect of the magnetic component of the electromagnetic field. It has also been shown that at high frequencies the nonlinear effects vanish, and the heating is well described by the linear theory. To verify the predicted new effect of plasma heating reduction at low frequencies, as well as to investigate the nonlinear effect of the magnetic field on plasma heating for arbitrary amplitudes of electromagnetic field in plasma, we have developed a 1d3v Particle-In-Cell (PIC) numerical simulation code with collisions. The collisions were implemented into the PIC code using two different techniques: the direct Monte-Carlo technique for the electron-atom collisions, and the stochastic technique based on the Langevin equation for the electron-electron collisions. The series of numerical simulations by this code confirmed the results of our linear theory, particularly the effect of heating reduction at low frequencies that we predicted theoretically. Also, the nonlinear effects of electromagnetic field on plasma heating were studied using the PIC code in the cases of weak and strong electromagnetic fields. It has been shown that in the case of weak electromagnetic fields (corresponding to weak nonlinearity) the nonlinear effects lead to some enhancement of heating (compared to the linear theory) at low frequencies, followed by a small reduction of heating at higher frequencies. This observed nonlinear perturbation of heating in warm plasma with collisions is similar to that predicted by the quasilinear theory for the case of cold plasma with collisions. In the case of strong electromagnetic fields (corresponding to strong nonlinearity) the nonlinear effects lead to a further reduction of heating (compared to the linear theory) at low frequencies, as shown by the simulation, thus adding to the effect of reduction of heating predicted by the linear theory. The nonlinear effects are shown to vanish at high frequencies, as expected.

In this thesis the nonlinear effects and heating are studied in inductively coupled plasma (ICP) in a regime of anomalous skin effect (nonlocal regime). In this regime the thermal motion of plasma electrons plays an important role, significantly influencing the processes associated with the penetration of electromagnetic field into plasma, such as the ponderomotive effect and heating of plasma by the field. We have developed a linear kinetic theory that describes the electron dynamics in ICP taking into account the electron thermal motion and collisions of electrons. This theory yields relatively simple expressions for the electron current in plasma, the ponderomotive force, and plasma heating. It describes correctly the thermal reduction of ponderomotive force in the nonlocal regime, which has been previously observed experimentally. It also describes the collisionless heating of plasma due to resonant interaction between the electromagnetic wave and plasma electrons. There is a good overall agreement of the results of our theory with the experimental data on ponderomotive force and plasma heating. Using our theory, we predicted a new effect of reduction of plasma heating compared to the purely collisional value, occurring at low frequencies. This effect has not been previously reported. The nonlinear effects of the electromagnetic field on the electron distribution function and on plasma heating, that are not accounted for in the linear kinetic theory, have been studied using a quasilinear kinetic theory, also developed in this thesis. Within the quasilinear approximation we have formulated the system of equations describing the slow response of plasma electrons to the fast oscillating electromagnetic field. As an example, these equations have been solved in the simplest case of cold plasma with collisions, and the nonlinear perturbation of the electron distribution function and its effect on the plasma heating have been found. It has been shown that the nonlinear modification of plasma heating occurs mainly due to the nonlinear effect of the magnetic component of the electromagnetic field. It has also been shown that at high frequencies the nonlinear effects vanish, and the heating is well described by the linear theory. To verify the predicted new effect of plasma heating reduction at low frequencies, as well as to investigate the nonlinear effect of the magnetic field on plasma heating for arbitrary amplitudes of electromagnetic field in plasma, we have developed a 1d3v Particle-In-Cell (PIC) numerical simulation code with collisions. The collisions were implemented into the PIC code using two different techniques: the direct Monte-Carlo technique for the electron-atom collisions, and the stochastic technique based on the Langevin equation for the electron-electron collisions. The series of numerical simulations by this code confirmed the results of our linear theory, particularly the effect of heating reduction at low frequencies that we predicted theoretically. Also, the nonlinear effects of electromagnetic field on plasma heating were studied using the PIC code in the cases of weak and strong electromagnetic fields. It has been shown that in the case of weak electromagnetic fields (corresponding to weak nonlinearity) the nonlinear effects lead to some enhancement of heating (compared to the linear theory) at low frequencies, followed by a small reduction of heating at higher frequencies. This observed nonlinear perturbation of heating in warm plasma with collisions is similar to that predicted by the quasilinear theory for the case of cold plasma with collisions. In the case of strong electromagnetic fields (corresponding to strong nonlinearity) the nonlinear effects lead to a further reduction of heating (compared to the linear theory) at low frequencies, as shown by the simulation, thus adding to the effect of reduction of heating predicted by the linear theory. The nonlinear effects are shown to vanish at high frequencies, as expected.

Children in Taiwan start their English classes in grade three. As they progress through the grades, they generally do not demonstrate high English proficiency. In addition, they also gradually lose their interest in learning English (“Education Minister,” 2009). To respond to the problem of achievement and motivation related to learning English, the Picture Word Inductive Model and Cooperative Learning were integrated as an alternative approach to the current instructional method in order to more effectively improve the Taiwanese elementary learners’ English literacy and highly motivate their interest in acquiring a foreign language. The focus of this study was to discover the participating teachers’ and students’ perspectives toward this new teaching approach and to understand the difficulties they encounter during the process of initiating and implementing an educational change. This eleven-month qualitative study involved three elementary English teachers and their 71 students from grades 4, 5 and 6 as participants. The data were collected through field notes from onsite classroom observations, teachers’ weekly reflective journals, in-class video recordings, and transcripts of teachers’ monthly meetings and personal interviews with the participants. Although one teacher and her students had to withdraw from the study because of intense pressure from school authorities and parents, the results of this study indicate that the remaining teachers and students highly recommended implementing this alternative approach in English classes and believed that this new way of teaching not only helped students become more autonomous and responsible for their own learning, but also provided them with more opportunities to interact with their peers. Although having doubts about this new approach at the beginning of the study and encountering difficulties during the process of implementation, the two teachers reported that their students’ English vocabulary had increased and they were able to compose meaningful English paragraphs as a result of this non-traditional strategy. The students also revealed that their motivation toward learning English had improved. Furthermore, the results show that support from school authorities and parents is essential to the initiation and maintenance of a change in education settings.

Concept discovery is the problem of finding definitions of target relation in terms or other relation given as a background knowledge. Inductive Logic Programming (ILP)-based and graph-based approaches are two competitors in concept discovery problem. Although ILP-based systems have long dominated the area, graph-based systems have recently gained popularity as they overcome certain shortcomings of ILP-based systems. While having applications in numerous domains, ILP-based concept discovery systems still sustain scalability and efficiency issues. These issues generally arose due to the large search spaces such systems build. In this work we propose memoization-based and parallelization-based methods that modify the search space construction step and the evaluation step of ILP-based concept discovery systems to overcome these problem. In this work we propose three memoization-based methods, called Tabular CRIS, Tabular CRIS-wEF, and Selective Tabular CRIS. In these methods, basically, evaluation queries are stored in look-up tables for later uses. While preserving some core functions in common, each proposed method improves e_ciency and scalability of its predecessor by introducing constraints on what kind of evaluation queries to store in look-up tables and for how long. The proposed parallelization method, called pCRIS, parallelizes the search space construction and evaluation steps of ILP-based concept discovery systems in a data-parallel manner. The proposed method introduces policies to minimize the redundant work and waiting time among the workers at synchronization points. Graph-based approaches were first introduced to the concept discovery domain to handle the so called local plateau problem. Graph-based approaches have recently gained more popularity in concept discovery system as they provide convenient environment to represent relational data and are able to overcome certain shortcomings of ILP-based concept discovery systems. Graph-based approaches can be classified as structure-based approaches and path-finding approaches. The first class of approaches need to employ expensive algorithms such as graph isomorphism to find frequently appearing substructures. The methods that fall into the second class need to employ sophisticated indexing mechanisms to find out the frequently appearing paths that connect some nodes in interest. In this work, we also propose a hybrid method for graph-based concept discovery which does not require costly substructure matching algorithms and path indexing mechanism. The proposed method builds the graph in such a way that similar facts are grouped together and paths that eventually turn to be concept descriptors are build while the graph is constructed.

Children in Taiwan start their English classes in grade three. As they progress through the grades, they generally do not demonstrate high English proficiency. In addition, they also gradually lose their interest in learning English (“Education Minister,” 2009). To respond to the problem of achievement and motivation related to learning English, the Picture Word Inductive Model and Cooperative Learning were integrated as an alternative approach to the current instructional method in order to more effectively improve the Taiwanese elementary learners’ English literacy and highly motivate their interest in acquiring a foreign language. The focus of this study was to discover the participating teachers’ and students’ perspectives toward this new teaching approach and to understand the difficulties they encounter during the process of initiating and implementing an educational change. This eleven-month qualitative study involved three elementary English teachers and their 71 students from grades 4, 5 and 6 as participants. The data were collected through field notes from onsite classroom observations, teachers’ weekly reflective journals, in-class video recordings, and transcripts of teachers’ monthly meetings and personal interviews with the participants. Although one teacher and her students had to withdraw from the study because of intense pressure from school authorities and parents, the results of this study indicate that the remaining teachers and students highly recommended implementing this alternative approach in English classes and believed that this new way of teaching not only helped students become more autonomous and responsible for their own learning, but also provided them with more opportunities to interact with their peers. Although having doubts about this new approach at the beginning of the study and encountering difficulties during the process of implementation, the two teachers reported that their students’ English vocabulary had increased and they were able to compose meaningful English paragraphs as a result of this non-traditional strategy. The students also revealed that their motivation toward learning English had improved. Furthermore, the results show that support from school authorities and parents is essential to the initiation and maintenance of a change in education settings.