Study of baleen whales’ ecology and interaction with maritime traffic activities to support management of a complex socio-ecological system

Thesis written in co-mentorship with Robert Michaud.

Adsorption energy and spin state of first-row transition metals adsorbed on MgO(100)

Slab and cluster model spin-polarized calculations have been carried out to study various properties of isolated first-row transition metal atoms adsorbed on the anionic sites of the regular MgO(100) surface. The calculated adsorption energies follow the trend of the metal cohesive energies, indicating that the changes in the metal-support and metal-metal interactions along the series are dominated by atomic properties. In all cases, except for Ni at the generalized gradient approximation level, the number of unpaired electron is maintained as in the isolated metal atom. The energy required to change the atomic state from high to low spin has been computed using the PW91 and B3LYP density-functional-theory-based methods. PW91 fails to predict the proper ground state of V and Ni, but the results for the isolated and adsorbed atom are consistent within the method. B3LYP properly predicts the ground state of all first-row transition atom the high- to low-spin transition considered is comparable to experiment. In all cases, the interaction with the surface results in a reduced high- to low-spin transition energy.

Comparison of different methods in estimating potential evapotranspiration at Muda Irrigation Scheme of Malaysia

Evapotranspiration (ET) is a complex process in the hydrological cycle that influences the quantity of runoff and thus the irrigation water requirements. Numerous methods have been developed to estimate potential evapotranspiration (PET). Unfortunately, most of the reliable PET methods are parameter rich models and therefore, not feasible for application in data scarce regions. On the other hand, accuracy and reliability of simple PET models vary widely according to regional climate conditions. The objective of the present study was to evaluate the performance of three temperature-based and three radiation-based simple ET methods in estimating historical ET and projecting future ET at Muda Irrigation Scheme at Kedah, Malaysia. The performance was measured by comparing those methods with the parameter intensive Penman-Monteith Method. It was found that radiation based methods gave better performance compared to temperature-based methods in estimation of ET in the study area. Future ET simulated from projected climate data obtained through statistical downscaling technique also showed that radiation-based methods can project closer ET values to that projected by Penman-Monteith Method. It is expected that the study will guide in selecting suitable methods for estimating and projecting ET in accordance to availability of meteorological data.

Alignment by Maximization of Mutual Information

A new information-theoretic approach is presented for finding the pose of an object in an image. The technique does not require information about the surface properties of the object, besides its shape, and is robust with respect to variations of illumination. In our derivation, few assumptions are made about the nature of the imaging process. As a result the algorithms are quite general and can foreseeably be used in a wide variety of imaging situations. Experiments are presented that demonstrate the approach registering magnetic resonance (MR) images with computed tomography (CT) images, aligning a complex 3D object model to real scenes including clutter and occlusion, tracking a human head in a video sequence and aligning a view-based 2D object model to real images. The method is based on a formulation of the mutual information between the model and the image called EMMA. As applied here the technique is intensity-based, rather than feature-based. It works well in domains where edge or gradient-magnitude based methods have difficulty, yet it is more robust than traditional correlation. Additionally, it has an efficient implementation that is based on stochastic approximation. Finally, we will describe a number of additional real-world applications that can be solved efficiently and reliably using EMMA. EMMA can be used in machine learning to find maximally informative projections of high-dimensional data. EMMA can also be used to detect and correct corruption in magnetic resonance images (MRI).

Gradient-based reinforcement learning techniques for underwater robotics behavior learning

Darrerament, l'interès pel desenvolupament d'aplicacions amb robots submarins autònoms (AUV) ha crescut de forma considerable. Els AUVs són atractius gràcies al seu tamany i el fet que no necessiten un operador humà per pilotar-los. Tot i això, és impossible comparar, en termes d'eficiència i flexibilitat, l'habilitat d'un pilot humà amb les escasses capacitats operatives que ofereixen els AUVs actuals. L'utilització de AUVs per cobrir grans àrees implica resoldre problemes complexos, especialment si es desitja que el nostre robot reaccioni en temps real a canvis sobtats en les condicions de treball. Per aquestes raons, el desenvolupament de sistemes de control autònom amb l'objectiu de millorar aquestes capacitats ha esdevingut una prioritat. Aquesta tesi tracta sobre el problema de la presa de decisions utilizant AUVs. El treball presentat es centra en l'estudi, disseny i aplicació de comportaments per a AUVs utilitzant tècniques d'aprenentatge per reforç (RL). La contribució principal d'aquesta tesi consisteix en l'aplicació de diverses tècniques de RL per tal de millorar l'autonomia dels robots submarins, amb l'objectiu final de demostrar la viabilitat d'aquests algoritmes per aprendre tasques submarines autònomes en temps real. En RL, el robot intenta maximitzar un reforç escalar obtingut com a conseqüència de la seva interacció amb l'entorn. L'objectiu és trobar una política òptima que relaciona tots els estats possibles amb les accions a executar per a cada estat que maximitzen la suma de reforços totals. Així, aquesta tesi investiga principalment dues tipologies d'algoritmes basats en RL: mètodes basats en funcions de valor (VF) i mètodes basats en el gradient (PG). Els resultats experimentals finals mostren el robot submarí Ictineu en una tasca autònoma real de seguiment de cables submarins. Per portar-la a terme, s'ha dissenyat un algoritme anomenat mètode d'Actor i Crític (AC), fruit de la fusió de mètodes VF amb tècniques de PG.

Development of molecular-based techniques for the detection, identification and quantification of food-borne pathogens

La presencia de microorganismos patógenos en alimentos es uno de los problemas esenciales en salud pública, y las enfermedades producidas por los mismos es una de las causas más importantes de enfermedad. Por tanto, la aplicación de controles microbiológicos dentro de los programas de aseguramiento de la calidad es una premisa para minimizar el riesgo de infección de los consumidores. Los métodos microbiológicos clásicos requieren, en general, el uso de pre-enriquecimientos no-selectivos, enriquecimientos selectivos, aislamiento en medios selectivos y la confirmación posterior usando pruebas basadas en la morfología, bioquímica y serología propias de cada uno de los microorganismos objeto de estudio. Por lo tanto, estos métodos son laboriosos, requieren un largo proceso para obtener resultados definitivos y, además, no siempre pueden realizarse. Para solucionar estos inconvenientes se han desarrollado diversas metodologías alternativas para la detección identificación y cuantificación de microorganismos patógenos de origen alimentario, entre las que destacan los métodos inmunológicos y moleculares. En esta última categoría, la técnica basada en la reacción en cadena de la polimerasa (PCR) se ha convertido en la técnica diagnóstica más popular en microbiología, y recientemente, la introducción de una mejora de ésta, la PCR a tiempo real, ha producido una segunda revolución en la metodología diagnóstica molecular, como pude observarse por el número creciente de publicaciones científicas y la aparición continua de nuevos kits comerciales. La PCR a tiempo real es una técnica altamente sensible -detección de hasta una molécula- que permite la cuantificación exacta de secuencias de ADN específicas de microorganismos patógenos de origen alimentario. Además, otras ventajas que favorecen su implantación potencial en laboratorios de análisis de alimentos son su rapidez, sencillez y el formato en tubo cerrado que puede evitar contaminaciones post-PCR y favorece la automatización y un alto rendimiento. En este trabajo se han desarrollado técnicas moleculares (PCR y NASBA) sensibles y fiables para la detección, identificación y cuantificación de bacterias patogénicas de origen alimentario (Listeria spp., Mycobacterium avium subsp. paratuberculosis y Salmonella spp.). En concreto, se han diseñado y optimizado métodos basados en la técnica de PCR a tiempo real para cada uno de estos agentes: L. monocytogenes, L. innocua, Listeria spp. M. avium subsp. paratuberculosis, y también se ha optimizado y evaluado en diferentes centros un método previamente desarrollado para Salmonella spp. Además, se ha diseñado y optimizado un método basado en la técnica NASBA para la detección específica de M. avium subsp. paratuberculosis. También se evaluó la aplicación potencial de la técnica NASBA para la detección específica de formas viables de este microorganismo. Todos los métodos presentaron una especificidad del 100 % con una sensibilidad adecuada para su aplicación potencial a muestras reales de alimentos. Además, se han desarrollado y evaluado procedimientos de preparación de las muestras en productos cárnicos, productos pesqueros, leche y agua. De esta manera se han desarrollado métodos basados en la PCR a tiempo real totalmente específicos y altamente sensibles para la determinación cuantitativa de L. monocytogenes en productos cárnicos y en salmón y productos derivados como el salmón ahumado y de M. avium subsp. paratuberculosis en muestras de agua y leche. Además este último método ha sido también aplicado para evaluar la presencia de este microorganismo en el intestino de pacientes con la enfermedad de Crohn's, a partir de biopsias obtenidas de colonoscopia de voluntarios afectados. En conclusión, este estudio presenta ensayos moleculares selectivos y sensibles para la detección de patógenos en alimentos (Listeria spp., Mycobacterium avium subsp. paratuberculosis) y para una rápida e inambigua identificación de Salmonella spp. La exactitud relativa de los ensayos ha sido excelente, si se comparan con los métodos microbiológicos de referencia y pueden serusados para la cuantificación de tanto ADN genómico como de suspensiones celulares. Por otro lado, la combinación con tratamientos de preamplificación ha resultado ser de gran eficiencia para el análisis de las bacterias objeto de estudio. Por tanto, pueden constituir una estrategia útil para la detección rápida y sensible de patógenos en alimentos y deberían ser una herramienta adicional al rango de herramientas diagnósticas disponibles para el estudio de patógenos de origen alimentario.

Model-based vehicle detection and classification using orthographic approximations

This paper reports the current state of work to simplify our previous model-based methods for visual tracking of vehicles for use in a real-time system intended to provide continuous monitoring and classification of traffic from a fixed camera on a busy multi-lane motorway. The main constraints of the system design were: (i) all low level processing to be carried out by low-cost auxiliary hardware, (ii) all 3-D reasoning to be carried out automatically off-line, at set-up time. The system developed uses three main stages: (i) pose and model hypothesis using 1-D templates, (ii) hypothesis tracking, and (iii) hypothesis verification, using 2-D templates. Stages (i) & (iii) have radically different computing performance and computational costs, and need to be carefully balanced for efficiency. Together, they provide an effective way to locate, track and classify vehicles.

View-based approaches to spatial representation in human vision

In an immersive virtual environment, observers fail to notice the expansion of a room around them and consequently make gross errors when comparing the size of objects. This result is difficult to explain if the visual system continuously generates a 3-D model of the scene based on known baseline information from interocular separation or proprioception as the observer walks. An alternative is that observers use view-based methods to guide their actions and to represent the spatial layout of the scene. In this case, they may have an expectation of the images they will receive but be insensitive to the rate at which images arrive as they walk. We describe the way in which the eye movement strategy of animals simplifies motion processing if their goal is to move towards a desired image and discuss dorsal and ventral stream processing of moving images in that context. Although many questions about view-based approaches to scene representation remain unanswered, the solutions are likely to be highly relevant to understanding biological 3-D vision.

Characterization and classification of tumor lesions using computerized fractal-based texture analysis and support vector machines in digital mammograms

Objective: This paper presents a detailed study of fractal-based methods for texture characterization of mammographic mass lesions and architectural distortion. The purpose of this study is to explore the use of fractal and lacunarity analysis for the characterization and classification of both tumor lesions and normal breast parenchyma in mammography. Materials and methods: We conducted comparative evaluations of five popular fractal dimension estimation methods for the characterization of the texture of mass lesions and architectural distortion. We applied the concept of lacunarity to the description of the spatial distribution of the pixel intensities in mammographic images. These methods were tested with a set of 57 breast masses and 60 normal breast parenchyma (dataset1), and with another set of 19 architectural distortions and 41 normal breast parenchyma (dataset2). Support vector machines (SVM) were used as a pattern classification method for tumor classification. Results: Experimental results showed that the fractal dimension of region of interest (ROIs) depicting mass lesions and architectural distortion was statistically significantly lower than that of normal breast parenchyma for all five methods. Receiver operating characteristic (ROC) analysis showed that fractional Brownian motion (FBM) method generated the highest area under ROC curve (A z = 0.839 for dataset1, 0.828 for dataset2, respectively) among five methods for both datasets. Lacunarity analysis showed that the ROIs depicting mass lesions and architectural distortion had higher lacunarities than those of ROIs depicting normal breast parenchyma. The combination of FBM fractal dimension and lacunarity yielded the highest A z value (0.903 and 0.875, respectively) than those based on single feature alone for both given datasets. The application of the SVM improved the performance of the fractal-based features in differentiating tumor lesions from normal breast parenchyma by generating higher A z value. Conclusion: FBM texture model is the most appropriate model for characterizing mammographic images due to self-affinity assumption of the method being a better approximation. Lacunarity is an effective counterpart measure of the fractal dimension in texture feature extraction in mammographic images. The classification results obtained in this work suggest that the SVM is an effective method with great potential for classification in mammographic image analysis.

A comparison between feature-based and EM-based contour tracking

Most active-contour methods are based either on maximizing the image contrast under the contour or on minimizing the sum of squared distances between contour and image 'features'. The Marginalized Likelihood Ratio (MLR) contour model uses a contrast-based measure of goodness-of-fit for the contour and thus falls into the first class. The point of departure from previous models consists in marginalizing this contrast measure over unmodelled shape variations. The MLR model naturally leads to the EM Contour algorithm, in which pose optimization is carried out by iterated least-squares, as in feature-based contour methods. The difference with respect to other feature-based algorithms is that the EM Contour algorithm minimizes squared distances from Bayes least-squares (marginalized) estimates of contour locations, rather than from 'strongest features' in the neighborhood of the contour. Within the framework of the MLR model, alternatives to the EM algorithm can also be derived: one of these alternatives is the empirical-information method. Tracking experiments demonstrate the robustness of pose estimates given by the MLR model, and support the theoretical expectation that the EM Contour algorithm is more robust than either feature-based methods or the empirical-information method. (c) 2005 Elsevier B.V. All rights reserved.

Sparse shape representation using the Laplace-Beltrami eigenfunctions and its application to modeling subcortical structures

We present a new sparse shape modeling framework on the Laplace-Beltrami (LB) eigenfunctions. Traditionally, the LB-eigenfunctions are used as a basis for intrinsically representing surface shapes by forming a Fourier series expansion. To reduce high frequency noise, only the first few terms are used in the expansion and higher frequency terms are simply thrown away. However, some lower frequency terms may not necessarily contribute significantly in reconstructing the surfaces. Motivated by this idea, we propose to filter out only the significant eigenfunctions by imposing l1-penalty. The new sparse framework can further avoid additional surface-based smoothing often used in the field. The proposed approach is applied in investigating the influence of age (38-79 years) and gender on amygdala and hippocampus shapes in the normal population. In addition, we show how the emotional response is related to the anatomy of the subcortical structures.

Cloud optical and microphysical properties derived from ground-based and satellite sensors over a site in the Yangtze Delta region

Comprehensive surface-based retrievals of cloud optical and microphysical properties were made at Taihu, a highly polluted site in the central Yangtze Delta region, during a research campaign from May 2008 to December 2009. Cloud optical depth (COD), effective radius (Re), and liquid water path (LWP) were retrieved from measurements made with a suite of ground-based and spaceborne instruments, including an Analytical Spectral Devices spectroradiometer, a multi␣lter rotating shadowband radiometer, a multichannel microwave radiometer profiler, and the Moderate Resolution Imaging Spectroradiometer (MODIS) on Terra and Aqua satellites. Retrievals from zenith radiance measurements capture better the temporal variation of cloud properties than do retrievals from hemispherical fluxes. Annual mean LWP, COD, and Re are 115.8 ± 90.8 g/m2, 28.5 ± 19.2, and 6.9 ± 4.2 microns. Over 90% of LWP values are less than 250 g/m2. Most of the COD values (>90%) fall between 5 and 60, and ~80% of Re values are less than 10 microns. Maximum (minimum) values of LWP and Re occur in summer (winter); COD is highest in winter and spring. Raining and nonraining clouds have signi␣cant differences in LWP, COD, and Re. Rainfall frequency is best correlated with LWP, followed by COD and Re. Cloud properties retrieved from multiple ground-based instruments are also compared with those from satellite retrievals. On average, relative to surface retrievals, mean differences of satellite retrievals in cloud LWP, COD, and Re were -33.6 g/m2 (-26.4%), -5.8 (-31.4%), and 2.9 ␣m (29.3%) for 11 MODIS-Terra overpasses and -43.3 g/m2 (-22.3%), -3.0 (-10.0%), and -1.3 ␣m (-12.0%) for 8 MODIS-Aqua overpasses, respectively. These discrepancies indicate that MODIS cloud products still suffer from large uncertainties in this region.

Software-based study of a non-sorting method for median calculation on a set of integer numbers

This paper presents a software-based study of a hardware-based non-sorting median calculation method on a set of integer numbers. The method divides the binary representation of each integer element in the set into bit slices in order to find the element located in the middle position. The method exhibits a linear complexity order and our analysis shows that the best performance in execution time is obtained when slices of 4-bit in size are used for 8-bit and 16-bit integers, in mostly any data set size. Results suggest that software implementation of bit slice method for median calculation outperforms sorting-based methods with increasing improvement for larger data set size. For data set sizes of N > 5, our simulations show an improvement of at least 40%.

Adsorption of organic molecules at the TiO2(110) surface: the effect of van der Waals interactions

Understanding the interaction of organic molecules with TiO2 surfaces is important for a wide range of technological applications. While density functional theory (DFT) calculations can provide valuable insight about these interactions, traditional DFT approaches with local exchange-correlation functionals suffer from a poor description of non-bonding van der Waals (vdW) interactions. We examine here the contribution of vdW forces to the interaction of small organic molecules (methane, methanol, formic acid and glycine) with the TiO2 (110) surface, based on DFT calculations with the optB88-vdW functional. The adsorption geometries and energies at different configurations were also obtained in the standard generalized gradient approximation (GGA-PBE) for comparison. We find that the optB88-vdW consistently gives shorter surface adsorbate-to-surface distances and slightly stronger interactions than PBE for the weak (physisorbed) modes of adsorption. In the case of strongly adsorbed (chemisorbed) molecules both functionals give similar results for the adsorption geometries, and also similar values of the relative energies between different chemisorption modes for each molecule. In particular both functionals predict that dissociative adsorption is more favourable than molecular adsorption for methanol, formic acid and glycine, in general agreement with experiment. The dissociation energies obtained from both functionals are also very similar, indicating that vdW interactions do not affect the thermodynamics of surface deprotonation. However, the optB88-vdW always predicts stronger adsorption than PBE. The comparison of the methanol adsorption energies with values obtained from a Redhead analysis of temperature programmed desorption data suggests that optB88-vdW significantly overestimates the adsorption strength, although we warn about the uncertainties involved in such comparisons.

Tensor regression based on linked multiway parameter analysis

Classical regression methods take vectors as covariates and estimate the corresponding vectors of regression parameters. When addressing regression problems on covariates of more complex form such as multi-dimensional arrays (i.e. tensors), traditional computational models can be severely compromised by ultrahigh dimensionality as well as complex structure. By exploiting the special structure of tensor covariates, the tensor regression model provides a promising solution to reduce the model’s dimensionality to a manageable level, thus leading to efficient estimation. Most of the existing tensor-based methods independently estimate each individual regression problem based on tensor decomposition which allows the simultaneous projections of an input tensor to more than one direction along each mode. As a matter of fact, multi-dimensional data are collected under the same or very similar conditions, so that data share some common latent components but can also have their own independent parameters for each regression task. Therefore, it is beneficial to analyse regression parameters among all the regressions in a linked way. In this paper, we propose a tensor regression model based on Tucker Decomposition, which identifies not only the common components of parameters across all the regression tasks, but also independent factors contributing to each particular regression task simultaneously. Under this paradigm, the number of independent parameters along each mode is constrained by a sparsity-preserving regulariser. Linked multiway parameter analysis and sparsity modeling further reduce the total number of parameters, with lower memory cost than their tensor-based counterparts. The effectiveness of the new method is demonstrated on real data sets.