FPGA implementation of a multi-population PBIL algorithm

Evolutionary-based algorithms play an important role in finding solutions to many problems that are not solved by classical methods, and particularly so for those cases where solutions lie within extreme non-convex multidimensional spaces. The intrinsic parallel structure of evolutionary algorithms are amenable to the simultaneous testing of multiple solutions; this has proved essential to the circumvention of local optima, and such robustness comes with high computational overhead, though custom digital processor use may reduce this cost. This paper presents a new implementation of an old, and almost forgotten, evolutionary algorithm: the population-based incremental learning method. We show that the structure of this algorithm is well suited to implementation within programmable logic, as compared with contemporary genetic algorithms. Further, the inherent concurrency of our FPGA implementation facilitates the integration and testing of micro-populations.

O sistema PBL, problem-based learning, no ensino de medicina no Brasil : análise bibliográfica sobre a sua execução

Introdução: Entre as estratégias de ensino e aprendizagem utilizadas nas práticas pedagógicas, a Problem Based Learning (PBL) (Aprendizagem Baseada em Problemas) é utilizada desde 1960, em especial nos cursos de Medicina. Mesmo sendo uma estratégia valiosa, um dos seus obstáculos é a pouca prática dos alunos em atividades autodirigidas, pesquisa e construção coletiva do conhecimento. Objetivo: Rastrear elementos constitutivos da PBL através de dados colhidos em artigos pesquisados em sítios de divulgação científica; Avaliar, nos estudos selecionados, os aspectos positivos e negativos que estejam relacionados com a metodologia do Sistema PBL aplicada ao ensino médico no Brasil. Metodologia: Estudo bibliográfico de 13 textos utilizando um modelo de desconstrução, denominada Análise Textual Discursiva (ATD) que consiste em: transformação dos artigos em pedaços menores; análise textual; identificação de padrões convergentes e divergentes em relação a PBL; organização e síntese dos dados, culminando com a elaboração de estratégia adaptativa da PBL para o curso de Medicina. Resultados: Foram encontradas 116 citações que convergiam para referências positivos acerca da metodologia PBL e 40 citações que divergiam acerca dos pontos positivos. Os aspectos positivos como o desenvolvimento de atitudes e habilidades; desenvolvimento de competências anteriores ao curso; efeitos positivos depois de terminada a graduação, como autonomia de estudo e a articulação entre currículo e realidade profissional, representam pontos a serem reforçados na aula. Em contraponto, foi observado que dentre os negativos a não compreensão do papel do professor como tutor; necessidade de conteúdo formal tradicional pelos alunos e a expectativa que o professor retire as suas dúvidas são pontos a serem evitados. Conclusões: A metodologia PBL deverá servir como metodologia ativa para aproveitar ao máximo as habilidades que os alunos já apresentam, potencializando o aprendizado na educação médica em sala de aula. Palavras-Chave: PBL; curso de medicina; metodologia ativa; educação médica.

Teaching College Writing in a High School Setting: The Impact of Concurrent Enrollment on Teacher Learning and Practice

Thesis (Ph.D.)--University of Washington, 2016-06

Optimization and Machine Learning Frameworks for Complex Network Analysis

Thesis (Ph.D.)--University of Washington, 2016-06

Using the EM Algorithm to Train Neural Networks: Misconceptions and a New Algorithm for Multiclass Classification

The expectation-maximization (EM) algorithm has been of considerable interest in recent years as the basis for various algorithms in application areas of neural networks such as pattern recognition. However, there exists some misconceptions concerning its application to neural networks. In this paper, we clarify these misconceptions and consider how the EM algorithm can be adopted to train multilayer perceptron (MLP) and mixture of experts (ME) networks in applications to multiclass classification. We identify some situations where the application of the EM algorithm to train MLP networks may be of limited value and discuss some ways of handling the difficulties. For ME networks, it is reported in the literature that networks trained by the EM algorithm using iteratively reweighted least squares (IRLS) algorithm in the inner loop of the M-step, often performed poorly in multiclass classification. However, we found that the convergence of the IRLS algorithm is stable and that the log likelihood is monotonic increasing when a learning rate smaller than one is adopted. Also, we propose the use of an expectation-conditional maximization (ECM) algorithm to train ME networks. Its performance is demonstrated to be superior to the IRLS algorithm on some simulated and real data sets.

From a Vicious Circle of Anxiety to a Virtuous Circle of Learning: Experience of Teaching to a Heterogenous Clientele

A considerable body of literature suggests that significant psychological barrier and anxiety characterize the teaching and learning process in statistics. This study investigates the incidence of statistics anxiety, the extent to which it can be overcome and the factors that contribute to the process of overcoming it. Self-study and overall teaching quality, amongst others, significantly contributed to this outcome. This study identifies factors contributing to overall teaching quality. The teaching and learning process typified a highly effective communication mechanism based on an appropriate diagnosis of individual needs. This cumulative change resulted from circular causation. It is argued that given appropriate conditions the vicious circle of anxiety can be transformed into a virtuous circle of learning.

Enhancing student performance through cooperative learning in physical sciences

Students in a physical sciences course were introduced to cooperative learning at the University of Queensland, Gatton Campus. Groups of four to five students worked together in tutorial and practical sessions. Mid-term and practical examinations were abolished and 40% of total marks were allocated to the cooperative learning activities. A peer- and self-assessment system was successfully adapted to account for individual performance in cooperative learning group assignments. The results suggest that cooperative learning was very well received by students, and they expressed willingness to join cooperative learning groups in other courses. In addition, cooperative learning offered many benefits to students in terms of graduate attributes such as teamwork, communication, lifelong learning and problem-solving.

GANN: Genetic algorithm neural networks for the detection of conserved combinations of features in DNA

Background: The multitude of motif detection algorithms developed to date have largely focused on the detection of patterns in primary sequence. Since sequence-dependent DNA structure and flexibility may also play a role in protein-DNA interactions, the simultaneous exploration of sequence-and structure-based hypotheses about the composition of binding sites and the ordering of features in a regulatory region should be considered as well. The consideration of structural features requires the development of new detection tools that can deal with data types other than primary sequence. Results: GANN ( available at http://bioinformatics.org.au/gann) is a machine learning tool for the detection of conserved features in DNA. The software suite contains programs to extract different regions of genomic DNA from flat files and convert these sequences to indices that reflect sequence and structural composition or the presence of specific protein binding sites. The machine learning component allows the classification of different types of sequences based on subsamples of these indices, and can identify the best combinations of indices and machine learning architecture for sequence discrimination. Another key feature of GANN is the replicated splitting of data into training and test sets, and the implementation of negative controls. In validation experiments, GANN successfully merged important sequence and structural features to yield good predictive models for synthetic and real regulatory regions. Conclusion: GANN is a flexible tool that can search through large sets of sequence and structural feature combinations to identify those that best characterize a set of sequences.

The effect of an algorithm-based sedation guideline on the duration of mechanical ventilation in an Australian intensive care unit

To examine the effect of an algorithm-based sedation guideline developed in a North American intensive care unit (ICU) on the duration of mechanical ventilation of patients in an Australian ICU. The intervention was tested in a pre-intervention, post-intervention comparative investigation in a 14-bed adult intensive care unit. Adult mechanically ventilated patients were selected consecutively (n =322) The pre-intervention and post-intervention groups were similar except for a higher number of patients with a neurological diagnosis in the pre-intervention group. An algorithm-based sedation guideline including a sedation scale was introduced using a multifaceted implementation strategy. The median duration of ventilation was 5.6 days in the post-intervention group, compared with 4.8 days for the pre-intervention group (P = 0.99). The length of stay was 8.2 days in the post-intervention group versus 7.1 days in the pre-intervention group (P = 0.04). There were no statistically significant differences for the other secondary outcomes, including the score on the Experience of Treatment in ICU 7 item questionnaire, number of tracheostomies and number of self-extubations. Records of compliance to recording the sedation score during both phases revealed that patients were slightly more deeply sedated when the guideline was used. The use of the algorithm-based sedation guideline did not reduce duration of mechanical ventilation in the setting of this study.

Convergence analysis of a discrete-time single-unit gradient ICA algorithm

We revisit the one-unit gradient ICA algorithm derived from the kurtosis function. By carefully studying properties of the stationary points of the discrete-time one-unit gradient ICA algorithm, with suitable condition on the learning rate, convergence can be proved. The condition on the learning rate helps alleviate the guesswork that accompanies the problem of choosing suitable learning rate in practical computation. These results may be useful to extract independent source signals on-line.

Copying actions and copying outcomes: Social learning through the second year

The present work documents how the logic of a model's demonstration and the communicative cues that the model provides interact with age to influence how children engage in social learning. Children at ages 12, 18, and 24 months (n = 204) watched a model open a series of boxes. Twelve-month-old subjects only copied the specific actions of the model when they were given a logical reason to do so- otherwise, they focused on reproducing the outcome of the demonstrated actions. Eighteen-month-old subjects focused on copying the outcome when the model was aloof. When the model acted socially, the subjects were as likely to focus on copying actions as outcomes, irrespective of the apparent logic of the model's behavior. Finally, 24-month-old subjects predominantly focused on copying the model's specific actions. However, they were less likely to produce the modeled outcome when the model acted nonsocially.

Incremental learning for interactive e-mail filtering

In this article, we propose a framework, namely, Prediction-Learning-Distillation (PLD) for interactive document classification and distilling misclassified documents. Whenever a user points out misclassified documents, the PLD learns from the mistakes and identifies the same mistakes from all other classified documents. The PLD then enforces this learning for future classifications. If the classifier fails to accept relevant documents or reject irrelevant documents on certain categories, then PLD will assign those documents as new positive/negative training instances. The classifier can then strengthen its weakness by learning from these new training instances. Our experiments’ results have demonstrated that the proposed algorithm can learn from user-identified misclassified documents, and then distil the rest successfully.

Sound source localisation through active audition

This paper presents a novel method for enabling a robot to determine the direction to a sound source through interacting with its environment. The method uses a new neural network, the Parameter-Less Self-Organizing Map algorithm, and reinforcement learning to achieve rapid and accurate response.

Sharing Narratives Online: An international, self-organising curriculum in rural health

The aim of the Rural Medicine Rotation (RMR) at the University of Queensland (UQ) is to give all third year medical students exposure to and an understanding of, clinical practice in Australian rural or remote locations. A difficulty in achieving this is the relatively short period of student clinical placements, in only one or two rural or remote locations. A web-based Clinical Discussion Board (CDB) has been introduced to address this problem by allowing students at various rural sites to discuss their rural experiences and clinical issues with each other. The rationale is to encourage an understanding of the breadth and depth of rural medicine through peer-based learning. Students are required to submit a minimum of four contributions over the course of their six week rural placement. Analysis of student usage patterns shows that the majority of students exceeded the minimum submission criteria indicating motivation rather than compulsion to contribute to the CDB. There is clear evidence that contributing or responding to the CDB develops studentâ??s critical thinking skills by giving and receiving assistance from peers, challenging attitudes and beliefs and stimulating reflective thought. This is particularly evident in regard to issues involving ethics or clinical uncertainty, subject areas that are not in the medical undergraduate curriculum, yet are integral to real-world medical practice. The CDB has proved to be a successful way to understand the concerns and interests of third year medical students immersed in their RMR and also in demonstrating how technology can help address the challenge of supporting students across large geographical areas. We have recently broadened this approach by including students from the Rural Program at The Ohio State University College of Medicine. This important international exchange of ideas and approaches to learning is expected to broaden clinical training content and improve understanding of rural issues.

Playing in continuous spaces: Some analysis and extension of population-based incremental learning

As an alternative to traditional evolutionary algorithms (EAs), population-based incremental learning (PBIL) maintains a probabilistic model of the best individual(s). Originally, PBIL was applied in binary search spaces. Recently, some work has been done to extend it to continuous spaces. In this paper, we review two such extensions of PBIL. An improved version of the PBIL based on Gaussian model is proposed that combines two main features: a new updating rule that takes into account all the individuals and their fitness values and a self-adaptive learning rate parameter. Furthermore, a new continuous PBIL employing a histogram probabilistic model is proposed. Some experiments results are presented that highlight the features of the new algorithms.