Space-time ocean wave measurement using variational stereo vision systems

Remote sensing imaging systems for the measurement of oceanic sea states have recently attracted renovated attention. Imaging technology is economical, non-invasive and enables a better understanding of the space-time dynamics of ocean waves over an area rather than at selected point locations of previous monitoring methods (buoys, wave gauges, etc.). We present recent progress in space-time measurement of ocean waves using stereo vision systems on offshore platforms. Both traditional disparity-based systems and modern elevation-based ones are presented in a variational optimization framework: the main idea is to pose the stereoscopic reconstruction problem of the surface of the ocean in a variational setting and design an energy functional whose minimizer is the desired temporal sequence of wave heights. The functional combines photometric observations as well as spatial and temporal smoothness priors. Disparity methods estimate the disparity between images as an intermediate step toward retrieving the depth of the waves with respect to the cameras, whereas elevation methods estimate the ocean surface displacements directly in 3-D space. Both techniques are used to measure ocean waves from real data collected at offshore platforms in the Black Sea (Crimean Peninsula, Ukraine) and the Northern Adriatic Sea (Venice coast, Italy). Then, the statistical and spectral properties of the resulting observed waves are analyzed. We show the advantages and disadvantages of the presented stereo vision systems and discuss the improvement of their performance in critical issues such as the robustness of the camera calibration in spite of undesired variations of the camera parameters.

Comportamiento del paciente de la clínica dental. Modelo basado en la percepción de la calidad del servicio y fidelización del paciente

Esta tesis doctoral propone un modelo de comportamiento del paciente de la clínica dental, basado en la percepción de la calidad del servicio (SERVQUAL), la fidelización del paciente, acciones de Marketing Relacional y aspectos socioeconómicos relevantes, de los pacientes de clínicas dentales. En particular, el estudio de campo se lleva a cabo en el ámbito geográfico de la Comunidad de Madrid, España, durante los años 2012 y 2013. La primera parte del proceso de elaboración del modelo está basada en la recolección de datos. Para ello, se realizaron cinco entrevistas a expertos dentistas y se aplicaron dos tipos encuestas diferentes: una para el universo formado por el conjunto de los pacientes de las clínicas dentales y la otra para el universo formado el conjunto de los dentistas de las clínicas dentales de la Comunidad de Madrid. Se obtuvo muestras de: 200 encuestas de pacientes y 220 encuestas de dentistas activos colegiados en el Ilustre Colegio Oficial de Odontólogos y Estomatólogos de la I Región Madrid. En la segunda parte de la elaboración del modelo, se realizó el análisis de los datos, la inducción y síntesis del modelo propuesto. Se utilizó la metodología de modelos gráficos probabilísticos, específicamente, una Red Bayesiana, donde se integraron variables (nodos) y sus dependencias estadísticas causales (arcos dirigidos), que representan el conocimiento obtenido de los datos recopilados en las encuestas y el conocimiento derivado de investigaciones precedentes en el área. Se obtuvo una Red Bayesiana compuesta por 6 nodos principales, de los cuales dos de ellos son nodos de observación directa: “Revisit Intention” y “SERVQUAL”, y los otros cuatro nodos restantes son submodelos (agrupaciones de variables), estos son respectivamente: “Attitudinal”, “Disease Information”, “Socioeconomical” y “Services”. Entre las conclusiones principales derivadas del uso del modelo, como herramientas de inferencia y los análisis de las entrevistas realizadas se obtiene que: (i) las variables del nodo “Attitudinal” (submodelo), son las más sensibles y significativas. Al realizarse imputaciones particulares en las variables que conforman el nodo “Attitudinal” (“RelationalMk”, “Satisfaction”, “Recommendation” y “Friendship”) se obtienen altas probabilidades a posteriori en la fidelidad del paciente de la clínica dental, medida por su intención de revisita. (ii) En el nodo “Disease Information” (submodelo) se destaca la relación de dependencia causal cuando se imputa la variable “Perception of disease” en “SERVQUAL”, demostrando que la percepción de la gravedad del paciente condiciona significativamente la percepción de la calidad del servicio del paciente. Como ejemplo destacado, si se realiza una imputación en la variable “Clinic_Type” se obtienen altas probabilidades a posteriori de las variables “SERVQUAL” y “Revisit Intention”, lo que evidencia, que el tipo de clínica dental influye significativamente en la percepción de la calidad del servicio y en la fidelidad del paciente (intención de revisita). (iii) En el nodo “Socioeconomical” (submodelo) la variable “Sex” resultó no ser significativa cuando se le imputaban diferentes valores, por el contrario, la variable “Age” e “Income” mostraban altas variabilidades en las probabilidades a posteriori cuando se imputaba alguna variable del submodelo “Services”, lo que evidencia, que estas variables condicionan la intención de contratar servicios (“Services”), sobretodo en las franjas de edad de 30 a 51 años en pacientes con ingresos entre 3000€ y 4000€. (iv) En el nodo “Services” (submodelo) los pacientes de las clínicas dentales mostraron altas probabilidades a priori para contratar servicios de fisiotrapia oral y gingival: “Dental Health Education” y “Parking”. (v) Las variables de fidelidad del paciente medidas desde su perspectiva comportamental que fueron utilizadas en el modelo: “Visit/year” “Time_clinic”, no aportaron información significativa. Tampoco, la variable de fidelidad del cliente (actitudinal): “Churn Efford”. (vi) De las entrevistas realizadas a expertos dentistas se obtiene que, los propietarios de la clínica tradicional tienen poca disposición a implementar nuevas estrategias comerciales, debido a la falta de formación en la gestión comercial y por falta de recursos y herramientas. Existe un rechazo generalizado hacia los nuevos modelos de negocios de clínicas dentales, especialmente en las franquicias y en lo que a políticas comerciales se refiere. Esto evidencia una carencia de gerencia empresarial en el sector. Como líneas futuras de investigación, se propone profundizar en algunas relaciones de dependencia (causales) como SERVQUALServices; SatisfactionServices; RelationalMKServices, Perception of diseaseSatisfaction, entre otras. Así como, otras variables de medición de la fidelidad comportamental que contribuyan a la mejora del modelo, como por ej. Gasto del paciente y rentabilidad de la visita. ABSTRACT This doctoral dissertation proposes a model of the behavior of the dental-clinic customer, based on the service-quality perception (SERVQUAL), loyalty, Relational Marketing and some relevant socio-economical characteristics, of the dental-clinic customers. In particular, the field study has been developed in the geographical region of Madrid, Spain during the years 2012 and 2013. The first stage of the preparation of the model consist in the data gathering process. For this purpose, five interviews where realized to expert dentists and also two different types of surveys: one for the universe defined by the set of dental-clinic patients and the second for the universe defined by the set of the dentists of the dental clinics of the Madrid Community. A sample of 200 surveys where collected for patients and a sample of 220 surveys where collected from active dentists belonging to the Ilustre Colegio Oficial de Odontólogos y Estomatólogos de la I Región Madrid. In the second stage of the model preparation, the processes of data-analysis, induction and synthesis of the final model where performed. The Graphic Probabilistic Models methodology was used to elaborate the final model, specifically, a Bayesian Network, where the variables (nodes) and their statistical and causal dependencies where integrated and modeled, representing thus, the obtained knowledge from the data obtained by the surveys and the scientific knowledge derived from previous research in the field. A Bayesian Net consisting on six principal nodes was obtained, of which two of them are directly observable: “Revisit Intention” y “SERVQUAL”, and the remaining four are submodels (a grouping of variables). These are: “Attitudinal”, “Disease Information”, “Socioeconomical” and “Services”. The main conclusions derived from the model, as an inference tool, and the analysis of the interviews are: (i) the variables inside the “Attitudinal” node are the most sensitive and significant. By making some particular imputations on the variables that conform the “Attitudinal” node (“RelationalMk”, “Satisfaction”, “Recommendation” y “Friendship”), high posterior probabilities (measured in revisit intention) are obtained for the loyalty of the dental-clinic patient. (ii) In the “Disease Information” node, the causal relation between the “Perception of disease” and “SERVQUAL” when “Perception of disease” is imputed is highlighted, showing that the perception of the severity of the patient’s disease conditions significantly the perception of service quality. As an example, by imputing some particular values to the “Clinic_Type” node high posterior probabilities are obtained for the “SERVQUAL” variables and for “Revisit Intention” showing that the clinic type influences significantly in the service quality perception and loyalty (revisit intention). (iii) In the “Socioeconomical” variable, the variable “Sex” showed to be non-significant, however, the “Age” variable and “Income” show high variability in its posterior probabilities when some variable from the “Services” node where imputed, showing thus, that these variables condition the intention to buy new services (“Services”), especially in the age range from 30 to 50 years in patients with incomes between 3000€ and 4000€. (iv) In the “Services” submodel the dental-clinic patients show high priors to buy services such as oral and gingival therapy, Dental Health Education and “Parking” service. (v) The obtained loyalty measures, from the behavioral perspective, “Visit/year” and “Time_clinic”, do not add significant information to the model. Neither the attitudinal loyalty component “Churn Efford”. (vi) From the interviews realized to the expert dentists it is observed that the owners of the traditional clinics have a low propensity to apply new commercial strategies due to a lack of resources and tools. In general, there exists an opposition to new business models in the sector, especially to the franchise dental model. All of this evidences a lack in business management in the sector. As future lines of research, a deep look into some statistical and causal relations is proposed, such as: SERVQUALServices; SatisfactionServices; RelationalMKServices, Perception of diseaseSatisfaction, as well as new measurement variables related to attitudinal loyalty that contribute to improve the model, for example, profit per patient and per visit.

Matchings and phylogenetic trees

This paper presents a natural coordinate system for phylogenetic trees using a correspondence with the set of perfect matchings in the complete graph. This correspondence produces a distance between phylogenetic trees, and a way of enumerating all trees in a minimal step order. It is useful in randomized algorithms because it enables moves on the space of trees that make random optimization strategies “mix” quickly. It also promises a generalization to intermediary trees when data are not decisive as to their choice of tree, and a new way of constructing Bayesian priors on tree space.

Bishop's transcripts at Wells /

Vol. 7 included in v. 5 of: North Petherton registers.

Assessing the impacts of grazing levels on bird density in woodland habitat: a Bayesian approach using expert opinion

Many studies on birds focus on the collection of data through an experimental design, suitable for investigation in a classical analysis of variance (ANOVA) framework. Although many findings are confirmed by one or more experts, expert information is rarely used in conjunction with the survey data to enhance the explanatory and predictive power of the model. We explore this neglected aspect of ecological modelling through a study on Australian woodland birds, focusing on the potential impact of different intensities of commercial cattle grazing on bird density in woodland habitat. We examine a number of Bayesian hierarchical random effects models, which cater for overdispersion and a high frequency of zeros in the data using WinBUGS and explore the variation between and within different grazing regimes and species. The impact and value of expert information is investigated through the inclusion of priors that reflect the experience of 20 experts in the field of bird responses to disturbance. Results indicate that expert information moderates the survey data, especially in situations where there are little or no data. When experts agreed, credible intervals for predictions were tightened considerably. When experts failed to agree, results were similar to those evaluated in the absence of expert information. Overall, we found that without expert opinion our knowledge was quite weak. The fact that the survey data is quite consistent, in general, with expert opinion shows that we do know something about birds and grazing and we could learn a lot faster if we used this approach more in ecology, where data are scarce. Copyright (c) 2005 John Wiley & Sons, Ltd.

Single patient (n-of-1) trials with binary treatment preference

The use of a fully parametric Bayesian method for analysing single patient trials based on the notion of treatment 'preference' is described. This Bayesian hierarchical modelling approach allows for full parameter uncertainty, use of prior information and the modelling of individual and patient sub-group structures. It provides updated probabilistic results for individual patients, and groups of patients with the same medical condition, as they are sequentially enrolled into individualized trials using the same medication alternatives. Two clinically interpretable criteria for determining a patient's response are detailed and illustrated using data from a previously published paper under two different prior information scenarios. Copyright (C) 2005 John Wiley & Sons, Ltd.

The population pharmacokinetics of citalopram after deliberate self-poisoning: A Bayesian approach

Defining the pharmacokinetics of drugs in overdose is complicated. Deliberate self-poisoning is generally impulsive and associated with poor accuracy in dose history. In addition, early blood samples are rarely collected to characterize the whole plasma-concentration time profile and the effect of decontamination on the pharmacokinetics is uncertain. The aim of this study was to explore a fully Bayesian methodology for population pharmacokinetic analysis of data that arose from deliberate self-poisoning with citalopram. Prior information on the pharmacokinetic parameters was elicited from 14 published studies on citalopram when taken in therapeutic doses. The data set included concentration-time data from 53 patients studied after 63 citalopram overdose events (dose range: 20-1700 mg). Activated charcoal was administered between 0.5 and 4 h after 17 overdose events. The clinical investigator graded the veracity of the patients' dosing history on a 5-point ordinal scale. Inclusion of informative priors stabilised the pharmacokinetic model and the population mean values could be estimated well. There were no indications of non-linear clearance after excessive doses. The final model included an estimated uncertainty of the dose amount which in a simulation study was shown to not affect the model's ability to characterise the effects of activated charcoal. The effect of activated charcoal on clearance and bioavailability was pronounced and resulted in a 72% increase and 22% decrease, respectively. These findings suggest charcoal administration is potentially beneficial after citalopram overdose. The methodology explored seems promising for exploring the dose-exposure relationship in the toxicological settings.

How to predict everything: Nostradamus in the role of Copernicus

Statistics is known to be an art as well as a science. The training of mathematical physicists predisposes them towards hypothesising plausible Bayesean priors. Tony Bracken and I were of that mind [1], but in our discussions we also recognised the Bayesean will-o'-the-wisp illustrated below.

Bayesian population pharmacokinetic analysis of sirolimus

Objective: It is usual that data collected from routine clinical care is sparse and unable to support the more complex pharmacokinetic (PK) models that may have been reported in previous rich data studies. Informative priors may be a pre-requisite for model development. The aim of this study was to estimate the population PK parameters of sirolimus using a fully Bayesian approach with informative priors. Methods: Informative priors including prior mean and precision of the prior mean were elicited from previous published studies using a meta-analytic technique. Precision of between-subject variability was determined by simulations from a Wishart distribution using MATLAB (version 6.5). Concentration-time data of sirolimus retrospectively collected from kidney transplant patients were analysed using WinBUGS (version 1.3). The candidate models were either one- or two-compartment with first order absorption and first order elimination. Model discrimination was based on computation of the posterior odds supporting the model. Results: A total of 315 concentration-time points were obtained from 25 patients. Most data were clustered at trough concentrations with range of 1.6 to 77 hours post-dose. Using informative priors, either a one- or two-compartment model could be used to describe the data. When a one-compartment model was applied, information was gained from the data for the value of apparent clearance (CL/F = 18.5 L/h), and apparent volume of distribution (V/F = 1406 L) but no information was gained about the absorption rate constant (ka). When a two-compartment model was fitted to the data, the data were informative about CL/F, apparent inter-compartmental clearance, and apparent volume of distribution of the peripheral compartment (13.2 L/h, 20.8 L/h, and 579 L, respectively). The posterior distribution of the volume distribution of central compartment and ka were the same as priors. The posterior odds for the two-compartment model was 8.1, indicating the data supported the two-compartment model. Conclusion: The use of informative priors supported the choice of a more complex and informative model that would otherwise have not been supported by the sparse data.

Precision-recall operating characteristic (P-ROC) curves in imprecise environments

Traditionally, machine learning algorithms have been evaluated in applications where assumptions can be reliably made about class priors and/or misclassification costs. In this paper, we consider the case of imprecise environments, where little may be known about these factors and they may well vary significantly when the system is applied. Specifically, the use of precision-recall analysis is investigated and compared to the more well known performance measures such as error-rate and the receiver operating characteristic (ROC). We argue that while ROC analysis is invariant to variations in class priors, this invariance in fact hides an important factor of the evaluation in imprecise environments. Therefore, we develop a generalised precision-recall analysis methodology in which variation due to prior class probabilities is incorporated into a multi-way analysis of variance (ANOVA). The increased sensitivity and reliability of this approach is demonstrated in a remote sensing application.

Bayesian invariant measurements of generalisation for discrete distributions

Neural network learning rules can be viewed as statistical estimators. They should be studied in Bayesian framework even if they are not Bayesian estimators. Generalisation should be measured by the divergence between the true distribution and the estimated distribution. Information divergences are invariant measurements of the divergence between two distributions. The posterior average information divergence is used to measure the generalisation ability of a network. The optimal estimators for multinomial distributions with Dirichlet priors are studied in detail. This confirms that the definition is compatible with intuition. The results also show that many commonly used methods can be put under this unified framework, by assume special priors and special divergences.

Regression with Gaussian processes

The Bayesian analysis of neural networks is difficult because the prior over functions has a complex form, leading to implementations that either make approximations or use Monte Carlo integration techniques. In this paper I investigate the use of Gaussian process priors over functions, which permit the predictive Bayesian analysis to be carried out exactly using matrix operations. The method has been tested on two challenging problems and has produced excellent results.

Gaussian processes for regression

The Bayesian analysis of neural networks is difficult because a simple prior over weights implies a complex prior distribution over functions. In this paper we investigate the use of Gaussian process priors over functions, which permit the predictive Bayesian analysis for fixed values of hyperparameters to be carried out exactly using matrix operations. Two methods, using optimization and averaging (via Hybrid Monte Carlo) over hyperparameters have been tested on a number of challenging problems and have produced excellent results.

Computing with infinite networks

For neural networks with a wide class of weight-priors, it can be shown that in the limit of an infinite number of hidden units the prior over functions tends to a Gaussian process. In this paper analytic forms are derived for the covariance function of the Gaussian processes corresponding to networks with sigmoidal and Gaussian hidden units. This allows predictions to be made efficiently using networks with an infinite number of hidden units, and shows that, somewhat paradoxically, it may be easier to compute with infinite networks than finite ones.

Prediction with Gaussian processes: from linear regression to linear prediction and beyond

The main aim of this paper is to provide a tutorial on regression with Gaussian processes. We start from Bayesian linear regression, and show how by a change of viewpoint one can see this method as a Gaussian process predictor based on priors over functions, rather than on priors over parameters. This leads in to a more general discussion of Gaussian processes in section 4. Section 5 deals with further issues, including hierarchical modelling and the setting of the parameters that control the Gaussian process, the covariance functions for neural network models and the use of Gaussian processes in classification problems.