Regression with input-dependent noise: A Gaussian process treatment

Gaussian processes provide natural non-parametric prior distributions over regression functions. In this paper we consider regression problems where there is noise on the output, and the variance of the noise depends on the inputs. If we assume that the noise is a smooth function of the inputs, then it is natural to model the noise variance using a second Gaussian process, in addition to the Gaussian process governing the noise-free output value. We show that prior uncertainty about the parameters controlling both processes can be handled and that the posterior distribution of the noise rate can be sampled from using Markov chain Monte Carlo methods. Our results on a synthetic data set give a posterior noise variance that well-approximates the true variance.

Recognition on Finite Set of Events: Bayesian Analysis of Generalization Ability and Classification Tree Pruning

The problem of recognition on finite set of events is considered. The generalization ability of classifiers for this problem is studied within the Bayesian approach. The method for non-uniform prior distribution specification on recognition tasks is suggested. It takes into account the assumed degree of intersection between classes. The results of the analysis are applied for pruning of classification trees.

Combining Information from Multiple Sources in Bayesian Modeling

Surveys can collect important data that inform policy decisions and drive social science research. Large government surveys collect information from the U.S. population on a wide range of topics, including demographics, education, employment, and lifestyle. Analysis of survey data presents unique challenges. In particular, one needs to account for missing data, for complex sampling designs, and for measurement error. Conceptually, a survey organization could spend lots of resources getting high-quality responses from a simple random sample, resulting in survey data that are easy to analyze. However, this scenario often is not realistic. To address these practical issues, survey organizations can leverage the information available from other sources of data. For example, in longitudinal studies that suffer from attrition, they can use the information from refreshment samples to correct for potential attrition bias. They can use information from known marginal distributions or survey design to improve inferences. They can use information from gold standard sources to correct for measurement error.

This thesis presents novel approaches to combining information from multiple sources that address the three problems described above.

The first method addresses nonignorable unit nonresponse and attrition in a panel survey with a refreshment sample. Panel surveys typically suffer from attrition, which can lead to biased inference when basing analysis only on cases that complete all waves of the panel. Unfortunately, the panel data alone cannot inform the extent of the bias due to attrition, so analysts must make strong and untestable assumptions about the missing data mechanism. Many panel studies also include refreshment samples, which are data collected from a random sample of new

individuals during some later wave of the panel. Refreshment samples offer information that can be utilized to correct for biases induced by nonignorable attrition while reducing reliance on strong assumptions about the attrition process. To date, these bias correction methods have not dealt with two key practical issues in panel studies: unit nonresponse in the initial wave of the panel and in the

refreshment sample itself. As we illustrate, nonignorable unit nonresponse

can significantly compromise the analyst's ability to use the refreshment samples for attrition bias correction. Thus, it is crucial for analysts to assess how sensitive their inferences---corrected for panel attrition---are to different assumptions about the nature of the unit nonresponse. We present an approach that facilitates such sensitivity analyses, both for suspected nonignorable unit nonresponse

in the initial wave and in the refreshment sample. We illustrate the approach using simulation studies and an analysis of data from the 2007-2008 Associated Press/Yahoo News election panel study.

The second method incorporates informative prior beliefs about

marginal probabilities into Bayesian latent class models for categorical data.

The basic idea is to append synthetic observations to the original data such that

(i) the empirical distributions of the desired margins match those of the prior beliefs, and (ii) the values of the remaining variables are left missing. The degree of prior uncertainty is controlled by the number of augmented records. Posterior inferences can be obtained via typical MCMC algorithms for latent class models, tailored to deal efficiently with the missing values in the concatenated data.

We illustrate the approach using a variety of simulations based on data from the American Community Survey, including an example of how augmented records can be used to fit latent class models to data from stratified samples.

The third method leverages the information from a gold standard survey to model reporting error. Survey data are subject to reporting error when respondents misunderstand the question or accidentally select the wrong response. Sometimes survey respondents knowingly select the wrong response, for example, by reporting a higher level of education than they actually have attained. We present an approach that allows an analyst to model reporting error by incorporating information from a gold standard survey. The analyst can specify various reporting error models and assess how sensitive their conclusions are to different assumptions about the reporting error process. We illustrate the approach using simulations based on data from the 1993 National Survey of College Graduates. We use the method to impute error-corrected educational attainments in the 2010 American Community Survey using the 2010 National Survey of College Graduates as the gold standard survey.

Modélisation incrémentale par méthode bayésienne

Les modèles incrémentaux sont des modèles statistiques qui ont été développés initialement dans le domaine du marketing. Ils sont composés de deux groupes, un groupe contrôle et un groupe traitement, tous deux comparés par rapport à une variable réponse binaire (le choix de réponses est « oui » ou « non »). Ces modèles ont pour but de détecter l’effet du traitement sur les individus à l’étude. Ces individus n’étant pas tous des clients, nous les appellerons : « prospects ». Cet effet peut être négatif, nul ou positif selon les caractéristiques des individus composants les différents groupes. Ce mémoire a pour objectif de comparer des modèles incrémentaux d’un point de vue bayésien et d’un point de vue fréquentiste. Les modèles incrémentaux utilisés en pratique sont ceux de Lo (2002) et de Lai (2004). Ils sont initialement réalisés d’un point de vue fréquentiste. Ainsi, dans ce mémoire, l’approche bayésienne est utilisée et comparée à l’approche fréquentiste. Les simulations sont e ectuées sur des données générées avec des régressions logistiques. Puis, les paramètres de ces régressions sont estimés avec des simulations Monte-Carlo dans l’approche bayésienne et comparés à ceux obtenus dans l’approche fréquentiste. L’estimation des paramètres a une influence directe sur la capacité du modèle à bien prédire l’effet du traitement sur les individus. Nous considérons l’utilisation de trois lois a priori pour l’estimation des paramètres de façon bayésienne. Elles sont choisies de manière à ce que les lois a priori soient non informatives. Les trois lois utilisées sont les suivantes : la loi bêta transformée, la loi Cauchy et la loi normale. Au cours de l’étude, nous remarquerons que les méthodes bayésiennes ont un réel impact positif sur le ciblage des individus composant les échantillons de petite taille.

Classifying Child Maltreatment by Brain Imaging Data

Thesis (Master's)--University of Washington, 2016-08

Bayesian Methods for Inferring Gene Regulatory Networks

Thesis (Ph.D.)--University of Washington, 2016-08

Modélisation incrémentale par méthode bayésienne

Les modèles incrémentaux sont des modèles statistiques qui ont été développés initialement dans le domaine du marketing. Ils sont composés de deux groupes, un groupe contrôle et un groupe traitement, tous deux comparés par rapport à une variable réponse binaire (le choix de réponses est « oui » ou « non »). Ces modèles ont pour but de détecter l’effet du traitement sur les individus à l’étude. Ces individus n’étant pas tous des clients, nous les appellerons : « prospects ». Cet effet peut être négatif, nul ou positif selon les caractéristiques des individus composants les différents groupes. Ce mémoire a pour objectif de comparer des modèles incrémentaux d’un point de vue bayésien et d’un point de vue fréquentiste. Les modèles incrémentaux utilisés en pratique sont ceux de Lo (2002) et de Lai (2004). Ils sont initialement réalisés d’un point de vue fréquentiste. Ainsi, dans ce mémoire, l’approche bayésienne est utilisée et comparée à l’approche fréquentiste. Les simulations sont e ectuées sur des données générées avec des régressions logistiques. Puis, les paramètres de ces régressions sont estimés avec des simulations Monte-Carlo dans l’approche bayésienne et comparés à ceux obtenus dans l’approche fréquentiste. L’estimation des paramètres a une influence directe sur la capacité du modèle à bien prédire l’effet du traitement sur les individus. Nous considérons l’utilisation de trois lois a priori pour l’estimation des paramètres de façon bayésienne. Elles sont choisies de manière à ce que les lois a priori soient non informatives. Les trois lois utilisées sont les suivantes : la loi bêta transformée, la loi Cauchy et la loi normale. Au cours de l’étude, nous remarquerons que les méthodes bayésiennes ont un réel impact positif sur le ciblage des individus composant les échantillons de petite taille.

Experimental separation of the coherent component of X-ray scattering prior to RDF analysis of non-crystalline polymers

An experimental method is described which enables the inelastically scattered X-ray component to be removed from diffractometer data prior to radial density function analysis. At each scattering angle an energy spectrum is generated from a Si(Li) detector combined with a multi-channel analyser from which the coherently scattered component is separated. The data obtained from organic polymers has an improved signal/noise ratio at high values of scattering angle, and a commensurate enhancement of resolution of the RDF at low r is demonstrated for the case of PMMA (ICI `Perspex'). The method obviates the need for the complicated correction for multiple scattering.

Joh. Helffrici Jungken ... Chymia experimentalis, sive, Naturalis philosophia mechanica : ubi prior pars generosiorum remediorum fabricam ex triplici regno cum omnibus manipulationibus, fideliter exhibet. Pars alter eadem medicamenta, inter alia, ad quoscunque morbos generaliter adaptare docet. Adjectis monitis medicis affectibus puerorum concernentibus. Nec non experimentis rerum naturalium principa commentarii, loco illustrantibus.

Mode of access: Internet.

Non-zero mean Gaussian process prior wind field models

This report outlines the derivation and application of a non-zero mean, polynomial-exponential covariance function based Gaussian process which forms the prior wind field model used in 'autonomous' disambiguation. It is principally used since the non-zero mean permits the computation of realistic local wind vector prior probabilities which are required when applying the scaled-likelihood trick, as the marginals of the full wind field prior. As the full prior is multi-variate normal, these marginals are very simple to compute.

[factors Associated With Satisfaction With Life Among Elderly Caregivers And Non-caregivers].

This article seeks to investigate associations between satisfaction with life and sociodemographic variables, health conditions, functionality, social involvement and social support among elderly caregivers and non-caregivers, as well as between satisfaction and the intensity of stress in the caregiver group. A sample of 338 caregivers was selected according to two items of the Brazilian version of the Elders Life Stress Inventory. A comparison-group of elderly non-caregivers was selected at random, with a similar gender, age and income profile. Data were derived from self-reported questionnaires and scales. Elderly caregivers with low levels of satisfaction and high levels of stress revealed more symptoms of insomnia, fatigue, diseases and worse IADL performance. Those with greater satisfaction and less stress revealed a good level of social support. Insomnia, depression and fatigue were associated with low satisfaction among caregivers, and with fatigue, depression and low social support among non-caregivers. It was considered relevant that instrumental, psychological and informative support can improve the quality of life and the quality of care provided by elderly caregivers, especially if they are affected by unfavorable health and psychosocial conditions and low satisfaction with life.

Microbial activity and community composition in saline and non-saline soils exposed to multiple drying and rewetting events

The effects of drying and rewetting (DRW) have been studied extensively in non-saline soils, but little is known about the impact of DRW in saline soils. An incubation experiment was conducted to determine the impact of 1-3 drying and re-wetting events on soil microbial activity and community composition at different levels of electrical conductivity in the saturated soil extract (ECe) (ECe 0.7, 9.3, 17.6 dS m(-1)). A non-saline sandy loam was amended with NaCl to achieve the three EC levels 21 days prior to the first DRW; wheat straw was added 7 days prior to the first DRW. Each DRW event consisted of 1 week drying and 1 week moist (50% of water holding capacity, WHC). After the last DRW, the soils were maintained moist until the end of the incubation period (63 days after addition of the wheat straw). A control was kept moist (50% of WHC) throughout the incubation period. Respiration rates on the day after rewetting were similar after the first and the second DRW, but significantly lower after the third DRW. After the first and second DRW, respiration rates were lower at EC17.6 compared to the lower EC levels, whereas salinity had little effect on respiration rates after the third DRW or at the end of the experiment when respiration rates were low. Compared to the continuously moist treatment, respiration rates were about 50% higher on day 15 (d15) and d29. On d44, respiration rates were about 50% higher at EC9.7 than at the other two EC levels. Cumulative respiration was increased by DRW only in the treatment with one DRW and only at the two lower EC levels. Salinity affected microbial biomass and community composition in the moist soils but not in the DRW treatments. At all EC levels and all sampling dates, the community composition in the continuously moist treatment differed from that in the DRW treatments, but there were no differences among the DRW treatments. Microbes in moderately saline soils may be able to utilise substrates released after multiple DRW events better than microbes in non-saline soil. However, at high EC (EC17.6), the low osmotic potential reduced microbial activity to such an extent that the microbes were not able to utilise substrate released after rewetting of dry soil.

Inhibition of settlement and metamorphosis of the ascidian Herdmania curvata by non-geniculate coralline algae

The surfaces of non-geniculate coralline algae (NCA) are known to induce the settlement and metamorphosis of disparate marine taxa. In this study we investigate the responsiveness of larvae of Herdmania curvata (Ascidiacea: Stolidobranchia) to three species of NCA (Neo-goniolithon brassica-florida, Hydrolithon onkodes, and Lithothamnium prolifer) that cohabit the slope and crest of Heron Reef, Great Barrier Reef. H. curvata larvae were first exposed to these NCA at or within 2 h of hatching, which is 1 to 2 h prior to attaining competence, and then cultured continuously with the NCA for 12 to 14 h. Rates of settlement and metamorphosis of H, curvata cultured in laboratory chambers in the presence of the different NCA were significantly lower than spontaneous rates in seawater. The limited settlement in treatments containing NCA were confined entirely to the chamber periphery, and settlement never occurred on the surface of the NCA. The inhibitory effect was dose-dependent and was stronger in H. brassica-florida and H. onkodes than in L. prolifer. Larvae that did not settle in treatments with NCA had rounded anterior trunks and, in extreme cases, kinked tails with rounded and dissociated tail muscle cells. In some individuals, we observed the anterior chemosensory papillae being sloughed off the larval body. Morphological analysis of trunk ectodermal and mesenchymal nuclei of larvae cultured in the presence of the NCA revealed that general necrotic cell death was occurring. Importantly, H. curvata larvae that were exposed to NCA could not subsequently be induced to metamorphose in KCl-elevated seawater, whereas larvae not exposed to NCA metamorphosed at high rates in KCl-elevated seawater.

Developmental vitamin D deficiency alters adult behaviour: An informative animal model of schizophrenia

Background: There is growing evidence that vitamin D is active in the brain but until recently there was a lack of evidence about its role during brain development. Guided by certain features of the epidemiology of schizophrenia, we have explored the role of vitamin D in the developing brain and behaviour using whole animal models. Methods: Sprague-Dawley rats were fed a vitamin D deficient diet (DVD) or control diet 6 weeks prior to mating and housed under UVB-free lighting conditions. On the day of birth all rats were fed a control diet for the remainder of the study. We observed behaviour at two timepoints; on the day of birth to study maternal behaviour, and at 10 weeks of age to study offspring behaviour in adulthood, under baseline and drug induced conditions (MK-801, haloperidol, amphetamine). Results: Prenatal vitamin D deficiency results in subtle alterations in maternal behaviour as well as long lasting effects on the adult offspring, despite a return to normal vitamin D levels during postnatal life. These affects were specific to transient prenatal vitamin D depletion as adult vitamin D depletion, combined prenatal and chronic postnatal vitamin D depletion, or ablation of the vitamin D receptor in mice led to markedly different outcomes. Conclusions: The developmental vitamin D (DVD) model now draws strength from epidemiological evidence of schizophrenia and animal experiments. Although the DVD model does not replicate every aspect of schizophrenia, it has several attractive features: (1) the exposure is based on clues from epidemiology; (2) it reproduces the increase in lateral ventricles; (3) it reproduces well-regarded behavioural phenotypes associated with schizophrenia (e.g. MK- 801 induced hyperlocomotion); and (4) it implicates a disturbance in dopamine signaling. In summary, low prenatal levels of vitamin D can influence critical components of orderly brain development and that this has a long lasting effect on behaviour.