Promedio bayesiano de modelos para la predicción del consumo máximo de oxígeno en deportistas con datos de no-ejercicio

Históricamente, los modelos de no-ejercicio para predecir el consumo máximo de oxígeno (VO2max) han sido construidos mediante regresión lineal frecuentista, usando técnicas estándar de selección de modelos. Sin embargo, existe incertidumbre acerca de la estructura estadística en el proceso de selección del modelo. En este estudio se propuso construir un modelo de no-ejercicio para predecir el VO2max en deportistas orientados al rendimiento, considerando la incertidumbre de modelo a través del Promedio Bayesiano de Modelos (BMA). Un objetivo adicional fue comparar la performance predictiva del BMA con las de los modelos derivados de varias técnicas frecuentistas usuales de selección de variables. Con tal fin, se implementó un submuestreo aleatorio estratificado repetido. Los datos incluyeron observaciones de la variable respuesta (en L·min-1), así como registros de Género, Deporte, Edad, Peso, Talla e Índice de masa corporal (BMI) (Edad = 22.1 ± 4.9 años, media ± SD; n = 272). Se propuso una clasificación de deportes con el objetivo de incluirla dentro del proceso de construcción del modelo: Combate, Juego, Resistencia 1 y Resistencia 2. El enfoque BMA se implementó en base a dos métodos: Occam's window y Composición de Modelo mediante el método de Monte Carlo con Cadenas de Markov (MC²). Se observaron discrepancias en la selección de variables entre los procedimientos frecuentistas. Ambos métodos de BMA produjeron resultados muy similares. Los modelos que incluyeron Género y las variables dummies para Resistencia 1 y Resistencia 2 acumularon virtualmente toda la probabilidad de modelo a posteriori. El Peso fue el predictor continuo con la más alta probabilidad de inclusión a posteriori (menor a 0.8). Las combinaciones de variables que involucraron predictores con un alto nivel de multicolinealidad fueron desacreditadas. Los modelos con sustancial contribución para el BMA presentaron un ajuste apreciable (R² ajustado menor a 0.8). Entre los modelos seleccionados por estrategias frecuentistas, el obtenido mediante el método de regresión por pasos (Stepwise regression method) con alfa igual a 0.05 fue el más respaldado por los datos, en términos de probabilidad de modelo a posteriori. En concordancia con la literatura, el BMA tuvo mejor performance predictiva de los datos fuera de la muestra que los modelos seleccionados por técnicas frecuentistas, medida por la cobertura del intervalo de predicción de 90 por ciento. La clasificación de deportes reveló resultados consistentes.

Alumnos recursantes de Botánica Agrícola I : un análisis descriptivo de esta situación

p.191-196

Places making: participatory construction of scientifi c culture cities

In the framework of the European project Platform of Local Authorities and Communicators Engaged in Science (PLACES), we analyse the articulations between scientifi c communication, public perception of science, processes of citizen participation and apropiation of space, based on a case study of the inhabitants of Teruel city, Autonomous Community of Aragon, Spain. On the interrelationships between these issues, there are a number of contradictions, such as the difference between a high interest for information about science and technology and a low level of recognition and interaction with local institutions involved in those activities, the complex conceptualization of scientifi c space in relation to the “public-private” pair, or an articulation of a claiming civic rethoric and an insuffi cient co-responsibility. We conclude that, in a local context, the dimension of territoriality and, in particular, the identifi cation with the town, is a central mediation for activating citizen participation as part of processes of appropriation of space for setting up cities of scientifi c culture.

A novel shrinkage technique based on the normal inverse Gaussian density model

This paper proposes a novel image denoising technique based on the normal inverse Gaussian (NIG) density model using an extended non-negative sparse coding (NNSC) algorithm proposed by us. This algorithm can converge to feature basis vectors, which behave in the locality and orientation in spatial and frequency domain. Here, we demonstrate that the NIG density provides a very good fitness to the non-negative sparse data. In the denoising process, by exploiting a NIG-based maximum a posteriori estimator (MAP) of an image corrupted by additive Gaussian noise, the noise can be reduced successfully. This shrinkage technique, also referred to as the NNSC shrinkage technique, is self-adaptive to the statistical properties of image data. This denoising method is evaluated by values of the normalized signal to noise rate (SNR). Experimental results show that the NNSC shrinkage approach is indeed efficient and effective in denoising. Otherwise, we also compare the effectiveness of the NNSC shrinkage method with methods of standard sparse coding shrinkage, wavelet-based shrinkage and the Wiener filter. The simulation results show that our method outperforms the three kinds of denoising approaches mentioned above.

Dietary patterns and bone mineral status in young adults: the Northern Ireland Young Hearts Project

Studies of individual nutrients or foods have revealed much about dietary influences on bone. Multiple food or nutrient approaches, such as dietary pattern analysis, could offer further insight but research is limited and largely confined to older adults. We examined the relationship between dietary patterns, obtained by a posteriori and a priori methods, and bone mineral status (BMS; collective term for bone mineral content (BMC) and bone mineral density (BMD)) in young adults (20-25 years; n 489). Diet was assessed by 7 d diet history and BMD and BMC were determined at the lumbar spine and femoral neck (FN). A posteriori dietary patterns were derived using principal component analysis (PCA) and three a priori dietary quality scores were applied (dietary diversity score (DDS), nutritional risk score and Mediterranean diet score). For the PCA-derived dietary patterns, women in the top compared to the bottom fifth of the 'Nuts and Meat' pattern had greater FN BMD by 0.074 g/cm(2) (P=0.049) and FN BMC by 0.40 g (P=0.034) after adjustment for confounders. Similarly, men in the top compared to the bottom fifth of the 'Refined' pattern had lower FN BMC by 0.41 g (P-0.049). For the a priori DDS, women in the top compared to the bottom third had lower FN BMD by 0.05 g/cm(2) after adjustments (P=0.052), but no other relationships with BMS were identified. In conclusion, adherence to a 'Nuts and Meat' dietary pattern may be associated with greater BMS in young women and a 'Refined' dietary pattern may be detrimental for bone health in young men.

Dietary patterns and cardiovascular risk factors in adolescents and young adults: the Northern Ireland Young Hearts Project

Dietary pattern (DP) analysis allows examination of the combined effects of nutrients and foods on the markers of CVD. Very few studies have examined these relationships during adolescence or young adulthood. Traditional CVD risk biomarkers were analysed in 12-15-year-olds (n 487; Young Hearts (YH)1) and again in the same individuals at 20-25 years of age (n 487; YH3). Based on 7 d diet histories, in the present study, DP analysis was performed using a posteriori principal component analysis for the YH3 cohort and the a priori Mediterranean Diet Score (MDS) was calculated for both YH1 and YH3 cohorts. In the a posteriori DP analysis, YH3 participants adhering most closely to the 'healthy' DP were found to have lower pulse wave velocity (PWV) and homocysteine concentrations, the 'sweet tooth' DP were found to have increased LDL concentrations, systolic blood pressure, and diastolic blood pressure and decreased HDL concentrations, the 'drinker/social' DP were found to have lower LDL and homocysteine concentrations, but exhibited a trend towards a higher TAG concentration, and finally the 'Western' DP were found to have elevated homocysteine and HDL concentrations. In the a priori dietary score analysis, YH3 participants adhering most closely to the Mediterranean diet were found to exhibit a trend towards a lower PWV. MDS did not track between YH1 and YH3, and nor was there a longitudinal relationship between the change in the MDS and the change in CVD risk biomarkers. In conclusion, cross-sectional analysis revealed that some associations between DP and CVD risk biomarkers were already evident in the young adult population, namely the association between the healthy DP (and the MDS) and PWV; however, no longitudinal associations were observed between these relatively short time periods.

Semi-Qualitative Probabilistic Networks in Computer Vision Problems

This paper explores the application of semi-qualitative probabilistic networks (SQPNs) that combine numeric and qualitative information to computer vision problems. Our version of SQPN allows qualitative influences and imprecise probability measures using intervals. We describe an Imprecise Dirichlet model for parameter learning and an iterative algorithm for evaluating posterior probabilities, maximum a posteriori and most probable explanations. Experiments on facial expression recognition and image segmentation problems are performed using real data.

New Complexity Results for MAP in Bayesian Networks

This paper presents new results for the (partial) maximum a posteriori (MAP) problem in Bayesian networks, which is the problem of querying the most probable state configuration of some of the network variables given evidence. It is demonstrated that the problem remains hard even in networks with very simple topology, such as binary polytrees and simple trees (including the Naive Bayes structure), which extends previous complexity results. Furthermore, a Fully Polynomial Time Approximation Scheme for MAP in networks with bounded treewidth and bounded number of states per variable is developed. Approximation schemes were thought to be impossible, but here it is shown otherwise under the assumptions just mentioned, which are adopted in most applications.

New Results for the MAP Problem in Bayesian Networks

This paper presents new results for the (partial) maximum a posteriori (MAP) problem in Bayesian networks, which is the problem of querying the most probable state configuration of some of the network variables given evidence. First, it is demonstrated that the problem remains hard even in networks with very simple topology, such as binary polytrees and simple trees (including the Naive Bayes structure). Such proofs extend previous complexity results for the problem. Inapproximability results are also derived in the case of trees if the number of states per variable is not bounded. Although the problem is shown to be hard and inapproximable even in very simple scenarios, a new exact algorithm is described that is empirically fast in networks of bounded treewidth and bounded number of states per variable. The same algorithm is used as basis of a Fully Polynomial Time Approximation Scheme for MAP under such assumptions. Approximation schemes were generally thought to be impossible for this problem, but we show otherwise for classes of networks that are important in practice. The algorithms are extensively tested using some well-known networks as well as random generated cases to show their effectiveness.

Improving parameter learning of Bayesian nets from incomplete data

This paper addresses the estimation of parameters of a Bayesian network from incomplete data. The task is usually tackled by running the Expectation-Maximization (EM) algorithm several times in order to obtain a high log-likelihood estimate. We argue that choosing the maximum log-likelihood estimate (as well as the maximum penalized log-likelihood and the maximum a posteriori estimate) has severe drawbacks, being affected both by overfitting and model uncertainty. Two ideas are discussed to overcome these issues: a maximum entropy approach and a Bayesian model averaging approach. Both ideas can be easily applied on top of EM, while the entropy idea can be also implemented in a more sophisticated way, through a dedicated non-linear solver. A vast set of experiments shows that these ideas produce significantly better estimates and inferences than the traditional and widely used maximum (penalized) log-likelihood and maximum a posteriori estimates. In particular, if EM is adopted as optimization engine, the model averaging approach is the best performing one; its performance is matched by the entropy approach when implemented using the non-linear solver. The results suggest that the applicability of these ideas is immediate (they are easy to implement and to integrate in currently available inference engines) and that they constitute a better way to learn Bayesian network parameters.

NP-hardness of MAP in Ternary Tree Bayesian Networks

This paper strengthens the NP-hardness result for the (partial) maximum a posteriori (MAP) problem in Bayesian networks with topology of trees (every variable has at most one parent) and variable cardinality at most three. MAP is the problem of querying the most probable state configuration of some (not necessarily all) of the network variables given evidence. It is demonstrated that the problem remains hard even in such simplistic networks.

The Inferential Complexity of Bayesian and Credal Networks

This paper presents new results on the complexity of graph-theoretical models that represent probabilities (Bayesian networks) and that represent interval and set valued probabilities (credal networks). We define a new class of networks with bounded width, and introduce a new decision problem for Bayesian networks, the maximin a posteriori. We present new links between the Bayesian and credal networks, and present new results both for Bayesian networks (most probable explanation with observations, maximin a posteriori) and for credal networks (bounds on probabilities a posteriori, most probable explanation with and without observations, maximum a posteriori).

Selection of burst-like transients and stochastic variables using multi-brand image differencing in the Pan-STARRS1 Medium Deep Survey

We present a novel method for the light-curve characterization of Pan-STARRS1 Medium Deep Survey (PS1 MDS) extragalactic sources into stochastic variables (SVs) and burst-like (BL) transients, using multi-band image-differencing time-series data. We select detections in difference images associated with galaxy hosts using a star/galaxy catalog extracted from the deep PS1 MDS stacked images, and adopt a maximum a posteriori formulation to model their difference-flux time-series in four Pan-STARRS1 photometric bands gP1, rP1, iP1, and zP1. We use three deterministic light-curve models to fit BL transients; a Gaussian, a Gamma distribution, and an analytic supernova (SN) model, and one stochastic light-curve model, the Ornstein-Uhlenbeck process, in order to fit variability that is characteristic of active galactic nuclei (AGNs). We assess the quality of fit of the models band-wise and source-wise, using their estimated leave-out-one cross-validation likelihoods and corrected Akaike information criteria. We then apply a K-means clustering algorithm on these statistics, to determine the source classification in each band. The final source classification is derived as a combination of the individual filter classifications, resulting in two measures of classification quality, from the averages across the photometric filters of (1) the classifications determined from the closest K-means cluster centers, and (2) the square distances from the clustering centers in the K-means clustering spaces. For a verification set of AGNs and SNe, we show that SV and BL occupy distinct regions in the plane constituted by these measures. We use our clustering method to characterize 4361 extragalactic image difference detected sources, in the first 2.5 yr of the PS1 MDS, into 1529 BL, and 2262 SV, with a purity of 95.00% for AGNs, and 90.97% for SN based on our verification sets. We combine our light-curve classifications with their nuclear or off-nuclear host galaxy offsets, to define a robust photometric sample of 1233 AGNs and 812 SNe. With these two samples, we characterize their variability and host galaxy properties, and identify simple photometric priors that would enable their real-time identification in future wide-field synoptic surveys.

The Complexity of MAP Inference in Bayesian Networks Specified Through Logical Languages

We study the computational complexity of finding maximum a posteriori configurations in Bayesian networks whose probabilities are specified by logical formulas. This approach leads to a fine grained study in which local information such as context-sensitive independence and determinism can be considered. It also allows us to characterize more precisely the jump from tractability to NP-hardness and beyond, and to consider the complexity introduced by evidence alone.

Contributo dos portfolios reflexivos no desenvolvimento profissional: um estudo de caso em contexto de formação complementar

Este estudo põe em evidência o valor formativo da reflexão pela escrita, estruturada em portfolios reflexivos, desenvolvidos durante a disciplina de Supervisão I, nas quatro turmas/cursos de formação complementar em Supervisão Pedagógica e Formação de Formadores, para o Exercício de Outras Funções, decorrida entre 1999-2004, na Escola Superior de Educação de Viseu, do Instituto Superior Politécnico de Viseu. Trata-se de uma investigação de índole qualitativa desenvolvida segundo a metodologia de estudo de caso, na vertente de estudo de casos múltiplos, dada a singularidade de que cada portfolio se reveste. A investigação foi perspectivada após a conclusão do referido curso com base na existência, ao tempo, de pouca investigação realizada no âmbito da utilização de portfolios na Formação Complementar de professores. As questões investigativas foram, por esse motivo, desenhadas a posteriori, o que podemos considerar uma limitação do estudo, ainda que os resultados obtidos confirmem as perspectivas desenvolvimentistas desta estratégia de formação, evidenciadas em investigações realizadas noutros contextos nacionais e internacionais. A adopção do portfolio reflexivo (Sá-Chaves, 2000) numa dupla dimensão – formativa e avaliativa – resultou de uma concepção de formação conducente ao desenvolvimento de professores reflexivos, capazes de exercer funções supervisivas alicerçadas na reflexão, no diálogo, na partilha e na intervenção. Pretendemos com a aplicação de portfolios reflexivos verificar até que ponto a construção de este tipo de narrativa autobiográfica podia contribuir para o desenvolvimento da profissionalidade docente de professores inseridos na carreira há já alguns/muitos anos, proporcionando-lhes a(s) ferramenta(s) reflexiva(s) necessária(s) a uma intervenção e/ou alteração de práticas mais adequadas às novas funções supervisivas que a legislação recente previa que viessem a exercer. Pudemos confirmar que a redacção do portfolio constituiu uma mais valia para os seus autores no que respeita ao seu desenvolvimento da sua profissionalidade, tendo-se verificado um crescimento evidente nas dimensões de conhecimento profissional, com especial incidência: no autoconhecimento; no conhecimento sobre estratégias reflexivas e supervisivas para o exercício de novas funções; no conhecimento de teorias práticas subjacentes às suas experiências supervisivas (passadas, presentes e futuras); no conhecimento sobre desenvolvimento curricular e sua flexibilização; no conhecimento sobre estratégias de avaliação alternativa e compreensiva; no conhecimento do Outro, no qual se incluem os alunos, os colegas de profissão e os restantes elementos da comunidade educativa; e no conhecimento sobre os contextos. Verificámos igualmente que a compreensão das competências supervisivas dos formandos também evoluiu. As reflexões vertidas neste estudo evidenciam que os formandos consciencializaram a necessidade de possuírem uma competência de intervenção fundamentada na reflexão, exercida sob uma perspectiva dialógica e humanista de supervisão e de adaptabilidade aos contextos profissionais. Acresce que alguns destes formandos manifestaram vontade e maior coragem para assumir essa intervenção, considerada imprescindível para que a escola evolua e se inove, como resposta à imprevisibilidade e à complexidade dos nossos tempos. Tornou-se igualmente evidente que os formandos desenvolveram as competências de integração da prática na teoria ou vice-versa e de autenticação das suas próprias teorias. Os professores são unânimes ao afirmar que a estratégia de portfolio contribuiu para o desenvolvimento da sua competência reflexiva e/ou de análise crítica, a qual, por sua vez, conduziu à auto-consciencialização da sua matriz identitária e à coragem para serem mais autênticos e interventivos/pró-activos. Terem sido capazes de ultrapassar as dificuldades e os receios colocados por uma estratégia que desconheciam, conferiu-lhes maior auto-estima e autonomia. Este aspecto está bem patente nas meta-análises realizadas por cada autor do portfolio, nas quais, aceitam a estratégia inicialmente combatida, reconhecendo-lhe valor formativo e a capacidade de revelação de conhecimento sobre si próprios. Pese embora o curto espaço de tempo em que a experiência de construção dos portfolios decorreu, tanto os formandos como nós consideramos que esta narrativa profissional autobiográfica é uma estratégia promotora de uma auto-supervisão contínua, que fornece e regista pistas de autodireccionamento profissional, sempre passíveis de avaliação e de redireccionamento. Com base nestas constatações consideramos que o portfolio reflexivo pode tornar-se no instrumento mais completo e diacrónico de auto-avaliação de professores, prevista no ECD, na medida em que permite retratar o desenvolvimento profissional e pessoal do seu autor e possibilita que este se reveja e redimensione a sua profissionalidade, sempre que acrescenta reflexões novas ou complementa anteriores, como resultado dos conselhos e/ou das discussões com o(s) Outro(s) e o próprio.