The space of compile-time transformations and or run-time options which can improve the performance of a given code is usually so large as to be virtually impossible to search in any practical time-frame. Thus, heuristics are leveraged which can suggest good but not necessarily best configurations. Unfortunately, since such heuristics are tightly coupled to processor architecture performance is not portable; heuristics must be tuned, traditionally manually, for each device in turn. This is extremely laborious and the result is often outdated heuristics and less effective optimisation. Ideally, to keep up with changes in hardware and run-time environments a fast and automated method to generate heuristics is needed. Recent works have shown that machine learning can be used to produce mathematical models or rules in their place, which is automated but not necessarily fast. This thesis proposes the use of active machine learning, sequential analysis, and active feature acquisition to accelerate the training process in an automatic way, thereby tackling this timely and substantive issue. First, a demonstration of the efficiency of active learning over the previously standard supervised machine learning technique is presented in the form of an ensemble algorithm. This algorithm learns a model capable of predicting the best processing device in a heterogeneous system to use per workload size, per kernel. Active machine learning is a methodology which is sensitive to the cost of training; specifically, it is able to reduce the time taken to construct a model by predicting how much is expected to be learnt from each new training instance and then only choosing to learn from those most profitable examples. The exemplar heuristic is constructed on average 4x faster than a baseline approach, whilst maintaining comparable quality. Next, a combination of active learning and sequential analysis is presented which reduces both the number of samples per training example as well as the number of training examples overall. This allows for the creation of models based on noisy information, sacrificing accuracy per training instance for speed, without having a significant affect on the quality of the final product. In particular, the runtime of high-performance compute kernels is predicted from code transformations one may want to apply using a heuristic which was generated up to 26x faster than with active learning alone. Finally, preliminary work demonstrates that an automated system can be created which optimises both the number of training examples as well as which features to select during training to further substantially accelerate learning, in cases where each feature value that is revealed comes at some cost.

No description available.

A área de aprendizado de máquina passa por uma grande expansão em seu universo de aplicações. Algoritmos de indução de modelos preditivos têm sido responsáveis pela realização de tarefas que eram inviáveis ou consideradas exclusividade do campo de ação humano até recentemente. Contudo, ainda é necessária a supervisão humana durante a construção de conjuntos de treinamento, como é o caso da tarefa de classificação. Tal construção se dá por meio da rotulação manual de cada exemplo, atribuindo a ele pelo menos uma classe. Esse processo, por ser manual, pode ter um custo elevado se for necessário muitas vezes. Uma técnica sob investigação corrente, capaz de mitigar custos de rotulação, é o aprendizado ativo. Dado um orçamento limitado, o objetivo de uma estratégia de amostragem ativa é direcionar o esforço de treinamento para os exemplos essenciais. Existem diversas abordagens efetivas de selecionar ativamente os exemplos mais importantes para consulta ao supervisor. Entretanto, não é possível, sem incorrer em custos adicionais, testá-las de antemão quanto à sua efetividade numa dada aplicação. Ainda mais crítica é a necessidade de que seja escolhido um algoritmo de aprendizado para integrar a estratégia de aprendizado ativo antes que se disponha de um conjunto de treinamento completo. Para lidar com esses desafios, esta tese apresenta como principais contribuições: uma estratégia baseada na inibição do algoritmo de aprendizado nos momentos menos propícios ao seu funcionamento; e, a experimentação da seleção de algoritmos de aprendizado, estratégias ativas de consulta ou pares estratégia-algoritmo baseada em meta-aprendizado, visando a experimentação de formas de escolha antes e durante o processo de rotulação. A estratégia de amostragem proposta é demonstrada competitiva empiricamente. Adicionalmente, experimentos iniciais com meta-aprendizado indicam a possibilidade de sua aplicação em aprendizado ativo, embora tenha sido identificado que investigações mais extensivas e aprofundadas sejam necessárias para apurar sua real efetividade prática. Importantes contribuições metodológicas são descritas neste documento, incluindo uma análise frequentemente negligenciada pela literatura da área: o risco devido à variabilidade dos algoritmos. Por fim, são propostas as curvas e faixas de ranqueamento, capazes de sumarizar, num único gráfico, experimentos de uma grande coleção de conjuntos de dados. / The machine learning area undergoes a major expansion in its universe of applications. Algorithms for the induction of predictive models have made it possible to carry out tasks that were once considered unfeasible or restricted to be solved by humans. However, human supervision is still needed to build training sets, for instance, in the classification task. Such building is usually performed by manual labeling of each instance, providing it, at least, one class. This process has a high cost due to its manual nature. A current technique under research, able to mitigate labeling costs, is called active learning. The goal of an active learning strategy is to manage the training effort to focus on the most relevant instances, within a budget. Several effective sampling approaches having been proposed. However, when one needs to choose the proper strategy for a given problem, they are impossible to test beforehand without incurring into additional costs. Even more critical is the need to choose a learning algorithm to integrate the active learning strategy before the existence of a complete training set. This thesis presents two major contributions to cope with such challenges: a strategy based on the learning algorithm inhibition when it is prone to inaccurate predictions; and, an attempt to automatically select the learning algorithms, active querying strategies or pairs strategy-algorithm, based on meta-learning. This attempt tries to verify the feasibility of such kind of decision making before and during the learning process. The proposed sampling approach is empirically shown to be competitive. Additionally, meta-learning experiments show that it can be applied to active learning, although more a extensive investigation is still needed to assess its real practical effectivity. Important methodological contributions are made in this document, including an often neglected analysis in the literature of active learning: the risk due to the algorithms variability. A major methodological contribution, called ranking curves, is presented.

We present INCREMENT, a cluster refinement algorithm which utilizes user feedback to refine clusterings. INCREMENT is capable of improving clusterings produced by arbitrary clustering algorithms. The initial clustering provided is first sub-clustered to improve query efficiency. A small set of select instances from each of these sub-clusters are presented to a user for labelling. Utilizing the user feedback, INCREMENT trains a feature embedder to map the input features to a new feature space. This space is learned such that spatial distance is inversely correlated with semantic similarity, determined from the user feedback. A final clustering is then formed in the embedded space. INCREMENT is tested on 9 datasets initially clustered with 4 distinct clustering algorithms. INCREMENT improved the accuracy of 71% of the initial clusterings with respect to a target clustering. For all the experiments the median percent improvement is 27.3% for V-Measure and is 6.08% for accuracy.

Thesis (M.Ed.) -- University of Limpopo, 2018 / This study investigated barriers to the reading of English texts in the rural schools of the Rakwadu Circuit in Mopani District, Limpopo Province. This problem is not only in the said Circuit, it is a world-wide challenge. The research was undertaken in three public secondary schools, with focus on the Grade 9 learners of the said Circuit. Data were collected through audio-taped interviews and observation of learners while reading prescribed texts. It was found that most educators and learners agree that there are barriers to the reading of English texts. The following factors were identified as barriers, namely: lack of libraries, non-parental involvement and insufficient learner-support materials, as well as lack of guided reading books. The study recommends that governmental officials should consider building libraries even in the rural schools and communities. The schools should have a parental involvement policy in which parents are encouraged to take part in the education of their children. The Department of Education should provide sufficient learner-support materials in schools to enhance learners’ reading ability. Educators are to be provided with guided reading materials that will enable them to implement Guided Reading approach. Curriculum advisors should train teachers on how to teach reading.

Objective. To document the type and extent of active-learning techniques used in US colleges and schools of pharmacy as well as factors associated with use of these techniques. Methods. A survey instrument was developed to assess whether and to what extent active learning was used by faculty members of US colleges and schools of pharmacy. This survey instrument was distributed via the American Association of Colleges of Pharmacy (AACP) mailing list. Results. Ninety-five percent (114) of all US colleges and schools of pharmacy were represented with at least 1 survey among the 1179 responses received. Eighty-seven percent of respondents used active-learning techniques in their classroom activities. The heavier the teaching workload the more active-learning strategies were used. Other factors correlated with higher use of active-learning strategies included younger faculty member age (inverse relationship), lower faculty member rank (inverse relationship), and departments that focused on practice, clinical and social, behavioral, and/or administrative sciences. Conclusions. Active learning has been embraced by pharmacy educators and is used to some extent by the majority of US colleges and schools of pharmacy. Future research should focus on how active-learning methods can be used most effectively within pharmacy education, how it can gain even broader acceptance throughout the academy, and how the effect of active learning on programmatic outcomes can be better documented.

Learning a predictive model for a large scale real-world problem presents several challenges: the choice of a good feature set and a scalable machine learning algorithm with small generalization error. A support vector machine (SVM), based on statistical learning theory, obtains good generalization by restricting the capacity of its hypothesis space. A SVM outperforms classical learning algorithms on many benchmark data sets. Its excellent performance makes it the ideal choice for pattern recognition problems. However, training a SVM involves constrained quadratic programming, which leads to poor scalability. In this dissertation, we propose several methods to improve a SVM's scalability. The evaluation is done mainly in the context of a plankton recognition problem. One approach is called active learning, which selectively asks a domain expert to label a subset of examples from a lot of unlabeled data. Active learning minimizes the number of labeled examples needed to build an accurate model and reduces the human effort in manually labeling the data. We propose a new active learning method "Breaking Ties" (BT) for multi-class SVMs. After developing a probability model for multiple class SVMs, "BT" selectively labels examples for which the difference in probabilities between the predicted most likely class and second most likely class is smallest. This simple strategy required several times less labeled plankton images to reach a given recognition accuracy when compared to random sampling in our plankton recognition system. To speed up a SVM's training and prediction, we show how to apply bit reduction to compress the examples into several bins. Weights are assigned to different bins based on the number of examples in the bin. Treating each bin as a weighted example, a SVM builds a model using the reduced-set of weighted examples.

This study's purpose was to investigate the differences in the attitudes towards science, belief in science, and the understanding of the nature of science between pre-service elementary education majors who took a two-semester interdisciplinary course called "Science That Matters" (ISC 1004 & ISC 1005) with those pre-service elementary education majors who took two undergraduate science courses other than the two-semester interdisciplinary science course. The research method employed a 30-item survey (Moore & Foy, 1997) entitled Scientific Attitude Inventory II. The survey's participants were two classes who had taken both semesters of the interdisciplinary course (n = 23) compared with six classes of elementary education majors who had taken two other undergraduate science courses other than the two-semester interdisciplinary course (n = 46). A two-tailed t-test was used to examine the differences in the means between the two groups as to their attitudes towards science, belief in science and their understanding of the nature of science. The study concluded that among the survey participants, there was no statistical difference as to the three dependent variables (attitudes towards science, belief in science, and the understanding of the nature of science) when testing for the independent variable (participants who had taken the two-semester interdisciplinary course and those who took two different science courses). The author suggests the results provide evidence that the two-semester interdisciplinary course holds its own when compared to other elective science courses, based on this evaluation of the students' attitudes toward science, belief in science and the understanding of the nature of science. Continuing research concerning this interdisciplinary course is needed to accumulate data which may show an advantage for students who take this course in learning and appreciating science for future elementary education teachers.

I have committed a significant period of time (in my case five years) to the purpose development of learning environments, with the belief that it would improve the self-actualisation and self-motivation of students and teachers alike. I consider it important to record and measure performance as we progressed toward such an outcome. Education researchers and practitioners alike, in the higher (university/tertiary) education systems, are seeking among new challenges to engage students and teachers in learning (James, 2001). However, studies to date show a confusing landscape littered with a multiplicity of interpretations and terms, successes and failures. As the discipline leader of the Information Technology, Systems and Multimedia (ITSM) Discipline, Swinburne University of Technology, Lilydale, I found myself struggling with this paradigm. I also found myself being torn between what presents as pragmatic student learning behaviour and the learner-centred teaching ideal reflected in the Swinburne Lilydale mission statement. The research reported in this folio reflects my theory and practice as discipline leader of the ITSM Discipline and the resulting learning environment evolution during the period 1997/8 to 2003. The study adds to the material evidence of extant research through firstly, a meta analysis of the learning environment implemented by the ITSM Discipline as recorded in peer reviewed and published papers; and secondly, a content analysis of student learning approaches, conducted on data reported from a survey of ‘learning skills inventory’ originally conducted by the ITSM Discipline staff in 2002. In 1997 information and communication technologies (ICT) were beginning to provide plausible means for electronic distribution of learning materials on a flexible and repeatable basis, and to provide answers to the imperative of learning materials distribution relating to an ITSM Discipline new course to begin in 1998. A very short time frame of three months was available prior to teaching the course. The ITSM Discipline learning environment development was an evolutionary process I began in 1997/8 initially from the requirement to publish print-based learning guide materials for the new ITSM Discipline subjects. Learning materials and student-to-teacher reciprocal communication would then be delivered and distributed online as virtual learning guides and virtual lectures, over distance as well as maintaining classroom-based instruction design. Virtual here is used to describe the use of ICT and Internet-based approaches. No longer would it be necessary for students to attend classes simply to access lecture content, or fear missing out on vital information. Assumptions I made as discipline leader for the ITSM Discipline included, firstly, that learning should be an active enterprise for the students, teachers and society; secondly, that each student comes to a learning environment with different learning expectations, learning skills and learning styles; and thirdly, that the provision of a holistic learning environment would encourage students to be self-actualising and self-motivated. Considerable reading of research and publications, as outlined in this folio, supported the update of these assumptions relative to teaching and learning. ITSM Discipline staff were required to quickly and naturally change their teaching styles and communication of values to engage with the emergent ITSM Discipline learning environment and pedagogy, and each new teaching situation. From a student perspective such assumptions meant students needed to move from reliance upon teaching and prescriptive transmission of information to a self-motivated and more self-actualising and reflective set of strategies for learning. In constructing this folio, after the introductory chaperts, there are two distinct component parts; • firstly, a Descriptive Meta analysis (Chapter Three) that draws together several of my peer reviewed professional writings and observations that document the progression of the ITSM Discipline learning environment evolution during the period 1997/8 to 2003. As the learning environment designer and discipline leader, my observations and published papers provide insight into the considerations that are required when providing an active, flexible and multi-modal learning environment for students and teachers; and • secondly, a Dissertation (Chapter Four), as a content analysis of a learning skills inventory data collection, collected by the ITSM Discipline in the 2002 Swinburne Lilydale academic year, where students were encouraged to complete reflective journal entries via the ITSM Discipline virtual learning guide subject web-site. That data collection included all students in a majority of subjects supported by the ITSM Discipline for both semesters one and two 2002. The original purpose of the journal entries was to have students reflectively involved in assessing their learning skills and approaches to learning. Such perceptions were tested using a well-known metric, the ‘learning skills inventory’ (Knowles, 1975), augmented with a short reflective learning approach narrative. The journal entries were used by teaching staff originally and then made available to researchers as a desensitised data in 2003 for statistical and content analysis relative to student learning skills and approaches. The findings of my research support a view of the student and teacher enculturation as utilitarian, dependent and pragmatically self-motivated. This, I argue, shows little sign of abatement in the early part of the 21st Century. My observation suggests that this is also independent of the pedagogical and educational philosophy debate or practice as currently presented. As much as the self-actualising, self-motivated learning environment can be justified philosophically, the findings observed from this research, reported in this folio, cannot. Part of the reason for this originates from the debate by educational researchers as to the relative merits of liberal and vocational philosophies for education combined with the recent introduction of information and communication technologies, and commodification of higher education. Challenging students to be participative and active learners, as proposed by educationalists Meyers and Jones (1993), i.e. self-motivated and self-actualising learners, has proved to be problematic. This, I will argue, will require a change to a variable/s (not yet identified) of higher education enculturation on multiple fronts, by students, teachers and society in order to bridge the gap. This research indicates that tertiary educators and educational researchers should stop thinking simplistically of constructivist and/or technology-enabled approaches, students learning choices and teachers teaching choices. Based on my research I argue for a far more holistic set of explanations of student and staff expectations and behaviour, and therefore pedagogy that supports those expectations.

This thesis presents a learning based approach for detecting classes of objects and patterns with variable image appearance but highly predictable image boundaries. It consists of two parts. In part one, we introduce our object and pattern detection approach using a concrete human face detection example. The approach first builds a distribution-based model of the target pattern class in an appropriate feature space to describe the target's variable image appearance. It then learns from examples a similarity measure for matching new patterns against the distribution-based target model. The approach makes few assumptions about the target pattern class and should therefore be fairly general, as long as the target class has predictable image boundaries. Because our object and pattern detection approach is very much learning-based, how well a system eventually performs depends heavily on the quality of training examples it receives. The second part of this thesis looks at how one can select high quality examples for function approximation learning tasks. We propose an {em active learning} formulation for function approximation, and show for three specific approximation function classes, that the active example selection strategy learns its target with fewer data samples than random sampling. We then simplify the original active learning formulation, and show how it leads to a tractable example selection paradigm, suitable for use in many object and pattern detection problems.