Understanding how the human visual system recognizes objects is one of the key challenges in neuroscience. Inspired by a large body of physiological evidence (Felleman and Van Essen, 1991; Hubel and Wiesel, 1962; Livingstone and Hubel, 1988; Tso et al., 2001; Zeki, 1993), a general class of recognition models has emerged which is based on a hierarchical organization of visual processing, with succeeding stages being sensitive to image features of increasing complexity (Hummel and Biederman, 1992; Riesenhuber and Poggio, 1999; Selfridge, 1959). However, these models appear to be incompatible with some well-known psychophysical results. Prominent among these are experiments investigating recognition impairments caused by vertical inversion of images, especially those of faces. It has been reported that faces that differ “featurally” are much easier to distinguish when inverted than those that differ “configurally” (Freire et al., 2000; Le Grand et al., 2001; Mondloch et al., 2002) – a finding that is difficult to reconcile with the aforementioned models. Here we show that after controlling for subjects’ expectations, there is no difference between “featurally” and “configurally” transformed faces in terms of inversion effect. This result reinforces the plausibility of simple hierarchical models of object representation and recognition in cortex.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Exploracionistas tem grande interesse em sistemas turbidíticos, pois em geral estes compõem prolíficos plays exploratórios. No entanto, estes potenciais reservatórios se encontram muitas vezes perto ou abaixo de resolução sísmica. Dessa forma, no processo de inversão, é importante integrar os dados sísmicos com qualidade de resolução obtida a partir do seu pré-condicionamento, para que o resultado final possua características detalhadas das camadas. O pré-condicionamento possibilita melhora na resolução dos dados sísmicos, através da atenuação dos ruídos aleatórios. Como objetivo final, foi realizada a inversão acústica em dados sísmicos post stack, migrados em tempo, a um sistema de turbiditos na Bacia de Campos. O principal objetivo da inversão sísmica é transformar o dado de reflexão em propriedades petrofísicas quantitativas. A inversão para impedância acústica é comumente utilizada para predição de porosidade. O fluxo de trabalho proposto foi dividido em cinco estágios principais: pré-condicionamento sísmico do dado 3D, correlação poço-sísmica, construção do modelo de baixa frequência, inversão do dado, e estimativa da porosidade. Os resultados mostraram que o cubo de impedância acústica invertido possui resolução muito superior quando comparado com o dado em amplitude sísmica, possibilitando melhor visualização das feições geológicas do Campo de Marlim. Além de suas limitações, como desconsiderando os efeitos das variações de fluido e variações litológicas complexas sobre a relação porosidade/impedância, o método fornece uma ferramenta confiável para exploração sísmica. Detalhes mais precisos das propriedades petrofísicas podem ser obtidos através de métodos de inversão mais sofisticados, a partir de dados pre stack. / Exploracionists has great interest in turbidite systems, since these generally comprise prolific exploration plays. However, these potential reservoirs are often near or below seismic resolution. Thus, the inversion process is important to integrate the seismic data quality of the resolution obtained from its preconditioning, so that the end result has detailed characteristics of the layers. The preconditioning enables improved resolution of seismic data, through mitigation of random noise. As the final goal, the inversion was performed acoustic post stack seismic data, migrated time, to a system of turbidites in Campos Basin. The main purpose of seismic inversion is to transform the given reflection in quantitative petrophysical properties. The inversion for acoustic impedance is commonly used to predict porosity. The flow of the proposed work was divided into five main stages: pre-conditioning the seismic 3D data, well-tie correlation, construction of the model of low frequency, inversion of data and estimation of porosity. The results showed that the inverted acoustic impedance cube has much higher resolution compared to the seismic amplitude data, allowing greater visualization of geological features in the Marlim Field. Besides its limitations as ignoring the effects of changing the fluid and complex lithological variations on the relationship porosity / impedance, the method provides a reliable tool for seismic exploration. More precise details of petrophysical properties can be obtainedthrough inversion methods starting from sophisticated pre-stack data.

This PhD thesis is primarily dedicated to a study of type III radio bursts observed by the S/Waves instrument onboard Solar TErrestrial RElations Observatory (STEREO). These emissions are produced by beams of suprathermal electrons escaping the corona along open magnetic field lines during increased solar activity. As fast electrons propagate in the interplanetary (IP) medium, Langmuir waves are generated at the local electron plasma frequency fpe by a bump-on-tail instability and can be afterwards converted by a non-linear process into radio emissions at fpe and/or 2fpe: type III radio bursts. We have developed a goniopolarimetric (GP, also referred to as direction-finding) inversion using the Singular Value Decomposition (SVD) technique for electric measurements on three non-orthogonal antennas. It allows us to retrieve both wave vector directions and polarization properties of incident waves. We have also investigated the influence of extended sources (as a typical feature of type III radio bursts) on measured spectral matrices. We have found an empirical relation between apparent source sizes and spectral matrices decomposed by SVD. Abovementioned techniques have been extensively tested on data obtained by the High Frequency Receiver (HFR, a part of S/Waves). We have performed statistical analysis of a...

Syftet med denna uppsats är att undersöka hur vuxna andraspråksinlärare hanterar svenskans omvända ordföljd eller inversion i påstående satser i skriftligt språk och att ta reda på hur de hanterar varierande placering av negationen i huvudsatser och bisatser. I undersökningen har tio skriftliga texter analyserats, samtliga skrivna av tio vuxna andraspråksinlärare i svenska som andraspråk 1. Undersökningen har både kvalitativa och kvantitativa inslag och utgår analysmässigt från ett lingvistiskt perspektiv samt från Pienemanns (1998) processbarhetsteori för andraspråksinlärning. Resultaten visar att majoriteten av inlärarna i en överväldigande majoritet av alla av icke-subjekt inledda påstående satser använder inversion målspråksenligt. Vad gäller placering av negationen klarar informanterna i hög grad av att göra detta målspråksenligt både i huvudsatser och bisatser, informanterna har dock svårigheter med placering av negationen i bisatser i hjälpverbskontext. I likhet med Bolanders (1987, 1988a, b, refererad i Philipsson, 2004) studie visar resultaten att informanterna behärskar negationen bättre i huvudsatser än bisatser.

Probabilistic ideas and tools have recently begun to permeate into several fields where they had traditionally not played a major role, including fields such as numerical linear algebra and optimization. One of the key ways in which these ideas influence these fields is via the development and analysis of randomized algorithms for solving standard and new problems of these fields. Such methods are typically easier to analyze, and often lead to faster and/or more scalable and versatile methods in practice. This thesis explores the design and analysis of new randomized iterative methods for solving linear systems and inverting matrices. The methods are based on a novel sketch-and-project framework. By sketching we mean, to start with a difficult problem and then randomly generate a simple problem that contains all the solutions of the original problem. After sketching the problem, we calculate the next iterate by projecting our current iterate onto the solution space of the sketched problem. The starting point for this thesis is the development of an archetype randomized method for solving linear systems. Our method has six different but equivalent interpretations: sketch-and-project, constrain-and-approximate, random intersect, random linear solve, random update and random fixed point. By varying its two parameters – a positive definite matrix (defining geometry), and a random matrix (sampled in an i.i.d. fashion in each iteration) – we recover a comprehensive array of well known algorithms as special cases, including the randomized Kaczmarz method, randomized Newton method, randomized coordinate descent method and random Gaussian pursuit. We also naturally obtain variants of all these methods using blocks and importance sampling. However, our method allows for a much wider selection of these two parameters, which leads to a number of new specific methods. We prove exponential convergence of the expected norm of the error in a single theorem, from which existing complexity results for known variants can be obtained. However, we also give an exact formula for the evolution of the expected iterates, which allows us to give lower bounds on the convergence rate. We then extend our problem to that of finding the projection of given vector onto the solution space of a linear system. For this we develop a new randomized iterative algorithm: stochastic dual ascent (SDA). The method is dual in nature, and iteratively solves the dual of the projection problem. The dual problem is a non-strongly concave quadratic maximization problem without constraints. In each iteration of SDA, a dual variable is updated by a carefully chosen point in a subspace spanned by the columns of a random matrix drawn independently from a fixed distribution. The distribution plays the role of a parameter of the method. Our complexity results hold for a wide family of distributions of random matrices, which opens the possibility to fine-tune the stochasticity of the method to particular applications. We prove that primal iterates associated with the dual process converge to the projection exponentially fast in expectation, and give a formula and an insightful lower bound for the convergence rate. We also prove that the same rate applies to dual function values, primal function values and the duality gap. Unlike traditional iterative methods, SDA converges under virtually no additional assumptions on the system (e.g., rank, diagonal dominance) beyond consistency. In fact, our lower bound improves as the rank of the system matrix drops. By mapping our dual algorithm to a primal process, we uncover that the SDA method is the dual method with respect to the sketch-and-project method from the previous chapter. Thus our new more general convergence results for SDA carry over to the sketch-and-project method and all its specializations (randomized Kaczmarz, randomized coordinate descent ... etc.). When our method specializes to a known algorithm, we either recover the best known rates, or improve upon them. Finally, we show that the framework can be applied to the distributed average consensus problem to obtain an array of new algorithms. The randomized gossip algorithm arises as a special case. In the final chapter, we extend our method for solving linear system to inverting matrices, and develop a family of methods with specialized variants that maintain symmetry or positive definiteness of the iterates. All the methods in the family converge globally and exponentially, with explicit rates. In special cases, we obtain stochastic block variants of several quasi-Newton updates, including bad Broyden (BB), good Broyden (GB), Powell-symmetric-Broyden (PSB), Davidon-Fletcher-Powell (DFP) and Broyden-Fletcher-Goldfarb-Shanno (BFGS). Ours are the first stochastic versions of these updates shown to converge to an inverse of a fixed matrix. Through a dual viewpoint we uncover a fundamental link between quasi-Newton updates and approximate inverse preconditioning. Further, we develop an adaptive variant of the randomized block BFGS (AdaRBFGS), where we modify the distribution underlying the stochasticity of the method throughout the iterative process to achieve faster convergence. By inverting several matrices from varied applications, we demonstrate that AdaRBFGS is highly competitive when compared to the well established Newton-Schulz and approximate preconditioning methods. In particular, on large-scale problems our method outperforms the standard methods by orders of magnitude. The development of efficient methods for estimating the inverse of very large matrices is a much needed tool for preconditioning and variable metric methods in the big data era.

Storm conditions are considered the dominating erosional mechanism for the coastal zone. Morphological changes during storms are hard to measure due to energetic environmental conditions. Surveys are therefore mostly executed right after a storm on a local scale over a single or few storms [days to weeks]. The impact of a single storm might depend on the preceding sequence of storms. Here, a video camera system is deployed in the South-West of England at the beach of Porthtowan to observe and assess short-term storm impact and long-term recovery. The morphological change is observed with a state-of-the-art video-based depth estimation tool that is based on the linear dispersion relationship between depth and wave celerity (cBathy). This work shows the first application of this depth estimation tool in a highly energetic macrotidal environment. Within this application two sources of first-order inaccuracies are identified: 1) camera related issues on the camera boundaries and 2) fixed pixel location for all tidal elevations. These systematic inaccuracies are overcome by 1) an adaptive pixel collection scheme and camera boundary solution and 2) freely moving pixels. The solutions together result in a maximum RMS-error reduction of 60%. From October 2013 to February 2015 depths are hourly estimated during daylight. This period included, the 2013-2014 winter season which was the most energetic winter since wave records began. Inter-tidal beach surveys show 200 m3/m erosion while the sub-tidal video derived bathymetries show a sediment loss of around 20 m3/m. At the same time the sub-tidal (outer) bar changes from 3D to linear due to a significant increase in alongshore wave power during storm conditions. Complex-EOF based storm-by-storm impact reveals that the individual storm impact at Porthtowan can be described as a combined function of storm-integrated incident offshore wave power [P] and disequilibrium and that the tidal range has limited effect on the storm impact. The inter- and sub-tidal domain together gain volume over the 2013-2014 winter and the two domains show an interactive inverse behaviour indicating sediment exchange during relatively calm summer conditions. The inter-tidal domain shows accelerated accretion during more energetic conditions in fall 2014. The outer bar slowly migrated onshore until more energetic wave conditions activate the sub-tidal storm deposits and 3 dimensionality is reintroduced. The inter-tidal beach shows full recovery in November 2014, 8 months after the stormy winter.

Le tuffeau est un matériau de construction calcaire largement utilisé dans la vallée de la Loire pour les bâtiments historiques. Le temps passant, il est confronté à des problèmes de pathologies, principalement liées à sa teneur en eau. Pour évaluer cet indicateur, le radar à sauts de fréquence (SFR) a été retenu parmi les techniques d’évaluation non destructives. Cette dernière peut être associée à une technique d’inversion « forme d’onde » dans le domaine fréquentiel (FWFI), technique performante pour obtenir la permittivité, directement en lien avec la teneur en eau. L’inversion FWFI d’un signal SFR utilise, pour le calcul direct, un modèle analytique de propagation des ondes, basé sur les fonctions de Green d’un dipôle, et sur le modèle de dispersion de Jonscher pour la caractérisation EM du milieu. Ce modèle direct est validé par une approche numérique utilisant le logiciel HFSS et par une approche expérimentale sur des blocs tests. Le problème inverse, comprenant une fonction Objectif et un algorithme de minimisation de type génétique, est ensuite étudié pour obtenir la permittivité du milieu ausculté. En parallèle, une base de données entre teneur en eau et permittivité du tuffeau est construite à l’aide de mesures EM en cellule sur des échantillons homogènes. Elle est associée à une analyse de régression de type vecteur-support pour prédire la teneur en eau du milieu à partir de la mesure de permittivité. Enfin, cette méthode est validée sur plusieurs applications pratiques, comprenant l’acquisition d’informations géométriques (estimation d’un front d’eau) et la caractérisation du matériau (estimation du gradient d’eau et cartographie hydrique). / Limestone is a historical building material widely used in Loire Valley of France. Time passed, it is facing to health problems, mainly due to water content. To evaluate this indicator, stepped-frequency radar is an efficient technique among all non-destructive testing techniques. It can be combined with a frequency-domain full-waveform inversion technique, which is an efficient signal processing method to achieve permittivity that is the observable of water content, after a calibration process. To do signal inversion, an analytical model of ultrawide band radar wave propagation through limestone is built in the forward study. This model is based on one-dipole Green’s function, for the EM propagation, and 4-parameters’ Jonscher dispersion model, for the EM characterization of the medium. This direct model is validated by a numerical approach using HFSS and an experimental approach using an experimental platform of Cerema. The inverse problem, including objective function and minimization algorithm, to achieve the permittivity, is then studied. An inversion method based on genetic algorithm is proposed and validated using numerical and experimental approaches. Afterwards, a large database between water content and permittvity of limestone is constructed using the electromagnetic cell test on cylindrical samples, based on which a support vector regression analysis is performed to predict water content using permittivity. Finally, this method is validated on several practical applications including geophysical information acquisition (water front estimation) and material characterization (water gradient estimation and hydric mapping).

The problem of fitting a material property of the earth to a certain model by analysing a returned seismic signal is investigated here. Analysis proceeds with methods taken from the theory of inverse problems. Seismic wave inversion is tack- led by minimisation of the objective function with respect to the model parameters. Absorbing boundary conditions are implemented using an exponentially decaying ansatz. / Physics / Ph. D. (Physics)

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Exploracionistas tem grande interesse em sistemas turbidíticos, pois em geral estes compõem prolíficos plays exploratórios. No entanto, estes potenciais reservatórios se encontram muitas vezes perto ou abaixo de resolução sísmica. Dessa forma, no processo de inversão, é importante integrar os dados sísmicos com qualidade de resolução obtida a partir do seu pré-condicionamento, para que o resultado final possua características detalhadas das camadas. O pré-condicionamento possibilita melhora na resolução dos dados sísmicos, através da atenuação dos ruídos aleatórios. Como objetivo final, foi realizada a inversão acústica em dados sísmicos post stack, migrados em tempo, a um sistema de turbiditos na Bacia de Campos. O principal objetivo da inversão sísmica é transformar o dado de reflexão em propriedades petrofísicas quantitativas. A inversão para impedância acústica é comumente utilizada para predição de porosidade. O fluxo de trabalho proposto foi dividido em cinco estágios principais: pré-condicionamento sísmico do dado 3D, correlação poço-sísmica, construção do modelo de baixa frequência, inversão do dado, e estimativa da porosidade. Os resultados mostraram que o cubo de impedância acústica invertido possui resolução muito superior quando comparado com o dado em amplitude sísmica, possibilitando melhor visualização das feições geológicas do Campo de Marlim. Além de suas limitações, como desconsiderando os efeitos das variações de fluido e variações litológicas complexas sobre a relação porosidade/impedância, o método fornece uma ferramenta confiável para exploração sísmica. Detalhes mais precisos das propriedades petrofísicas podem ser obtidos através de métodos de inversão mais sofisticados, a partir de dados pre stack. / Exploracionists has great interest in turbidite systems, since these generally comprise prolific exploration plays. However, these potential reservoirs are often near or below seismic resolution. Thus, the inversion process is important to integrate the seismic data quality of the resolution obtained from its preconditioning, so that the end result has detailed characteristics of the layers. The preconditioning enables improved resolution of seismic data, through mitigation of random noise. As the final goal, the inversion was performed acoustic post stack seismic data, migrated time, to a system of turbidites in Campos Basin. The main purpose of seismic inversion is to transform the given reflection in quantitative petrophysical properties. The inversion for acoustic impedance is commonly used to predict porosity. The flow of the proposed work was divided into five main stages: pre-conditioning the seismic 3D data, well-tie correlation, construction of the model of low frequency, inversion of data and estimation of porosity. The results showed that the inverted acoustic impedance cube has much higher resolution compared to the seismic amplitude data, allowing greater visualization of geological features in the Marlim Field. Besides its limitations as ignoring the effects of changing the fluid and complex lithological variations on the relationship porosity / impedance, the method provides a reliable tool for seismic exploration. More precise details of petrophysical properties can be obtainedthrough inversion methods starting from sophisticated pre-stack data.

Neste estudo é apresentado um procedimento de análise de dados magnéticos em casos de presença de magnetização remanente para, ao final do processo, a realização de uma inversão mais rápida e fiel ao comportamento do alvo em sub-superfície. O procedimento é composto por seis passos: (1) Delimitação lateral do alvo; (2) Estimativa de suas profundidades; (3) Estimativa das direções de inclinação e declinação aparentes de seu vetor de magnetização total; (4) Modelagem inicial com inferência de valores de susceptibilidades oriundos de estudos geológicos anteriores sobre o alvo e/ou sua região; (5) Inversão dos dados magnéticos utilizando o modelo inicial; e (6) atribuição das características magnéticas do passo (3) sobre o modelo de contraste de susceptibilidade magnética obtido com a inversão para a modelagem final do alvo, conhecendo-se seus vetores de magnetização induzida, total e, por subtração vetorial de ambos, remanente. Este procedimento foi aplicado a 108 casos sintéticos e a 8 casos reais pertencentes às províncias ígneas do Alto do Paranaíba e Rondoniana-San-Ignácio. Os resultados do uso deste procedimento indicaram uma recuperação das direções das componentes de magnetização com erro menor que 10%, em casos sintéticos, uma redução de mais de 20% no tempo de inversão com o uso de modelos iniciais, e qualitativamente, apresentaram modelos mais próximos dos originais (nos casos sintéticos) e geologicamente factíveis nos casos reais. / In this study, we present a procedure of analysis of magnetic data when remanence is present in order to, at the end of the process, obtain an inversion faster and more reliable inversion. The procedure is composed of six steps: (1) Estimation of the borders of the target; (2) Estimation of its depths; (3) A sweeping for the total apparent inclination and declination directions; (4) Initial modeling of a synthetic body, based on the recovered geometry and depth, on the directions of inclination and declination of the total magnetic field, and on previous analysis of the target and/or its region; (5) Inversion of magnetic data using the initial model; and (6) Attribution of the magnetic features of step (3) to the model recovered by the inversion for a final modeling of the target, estimating as well its remanent magnetization; the last through the vectorial resultant of the induction and total magnetization subtraction. This procedure was applied to 108 synthetic and to 8 real cases from the Alto do Paranaíba and Rondonian-San Ignacio Igneous Provinces. Their results pointed out that the error between the recovered directions of the magnetization components and the original values, in synthetic cases, was smaller than 10%; The inversions had their processing-time reduced in more than 20% and, qualitatively, presented models were more similar to the original (synthetic cases) and geologically feasible (real cases).