Theese objects from burned clay are representing the sublimation of traditional technologies and materials with contemporary visual technologies such as 3D.

Cette thèse présente nos contributions àl’inférence et l’apprentissage des modèles graphiquesen vision artificielle. Tout d’abord, nous proposons unenouvelle classe d’algorithmes de décomposition pour résoudrele problème d’appariement de graphes et d’hypergraphes,s’appuyant sur l’algorithme des directionsalternées (ADMM) non convexe. Ces algorithmes sontefficaces en terme de calcul et sont hautement parallélisables.En outre, ils sont également très générauxet peuvent être appliqués à des fonctionnelles d’énergiearbitraires ainsi qu’à des contraintes de correspondancearbitraires. Les expériences montrent qu’ils surpassentles méthodes de pointe existantes sur des benchmarkspopulaires. Ensuite, nous proposons une relaxationcontinue non convexe pour le problème d’estimationdu maximum a posteriori (MAP) dans les champsaléatoires de Markov (MRFs). Nous démontrons quecette relaxation est serrée, c’est-à-dire qu’elle est équivalenteau problème original. Cela nous permet d’appliquerdes méthodes d’optimisation continue pour résoudrele problème initial discret sans perte de précisionaprès arrondissement. Nous étudions deux méthodes degradient populaires, et proposons en outre une solutionplus efficace utilisant l’ADMM non convexe. Les expériencessur plusieurs problèmes réels démontrent quenotre algorithme prend l’avantage sur ceux de pointe,dans différentes configurations. Finalement, nous proposonsune méthode d’apprentissage des paramètres deces modèles graphiques avec des données d’entraînement,basée sur l’ADMM non convexe. Cette méthodeconsiste à visualiser les itérations de l’ADMM commeune séquence d’opérations différenciables, ce qui permetde calculer efficacement le gradient de la perted’apprentissage par rapport aux paramètres du modèle.L’apprentissage peut alors utiliser une descente de gradientstochastique. Nous obtenons donc un frameworkunifié pour l’inférence et l’apprentissage avec l’ADMMnon-convexe. Grâce à sa flexibilité, ce framework permetégalement d’entraîner conjointement de-bout-en-boutun modèle graphique avec un autre modèle, telqu’un réseau de neurones, combinant ainsi les avantagesdes deux. Nous présentons des expériences sur un jeude données de segmentation sémantique populaire, démontrantl’efficacité de notre méthode. / This thesis presents our contributions toinference and learning of graph-based models in computervision. First, we propose a novel class of decompositionalgorithms for solving graph and hypergraphmatching based on the nonconvex alternating directionmethod of multipliers (ADMM). These algorithms arecomputationally efficient and highly parallelizable. Furthermore,they are also very general and can be appliedto arbitrary energy functions as well as arbitraryassignment constraints. Experiments show that theyoutperform existing state-of-the-art methods on popularbenchmarks. Second, we propose a nonconvex continuousrelaxation of maximum a posteriori (MAP) inferencein discrete Markov random fields (MRFs). Weshow that this relaxation is tight for arbitrary MRFs.This allows us to apply continuous optimization techniquesto solve the original discrete problem withoutloss in accuracy after rounding. We study two populargradient-based methods, and further propose a more effectivesolution using nonconvex ADMM. Experimentson different real-world problems demonstrate that theproposed ADMM compares favorably with state-of-theartalgorithms in different settings. Finally, we proposea method for learning the parameters of these graphbasedmodels from training data, based on nonconvexADMM. This method consists of viewing ADMM iterationsas a sequence of differentiable operations, whichallows efficient computation of the gradient of the trainingloss with respect to the model parameters, enablingefficient training using stochastic gradient descent. Atthe end we obtain a unified framework for inference andlearning with nonconvex ADMM. Thanks to its flexibility,this framework also allows training jointly endto-end a graph-based model with another model suchas a neural network, thus combining the strengths ofboth. We present experiments on a popular semanticsegmentation dataset, demonstrating the effectivenessof our method.

Nous étudions les relations qui existent entre des aspects de la géométrie sous-riemannienne et une diversité de singularités typiques dans ce contexte.Avec les théorèmes de Whitney sous-riemanniens, nous conditionnons l’existence de prolongements globaux de courbes horizontales définies sur des fermés à des hypothèses de non-singularité de l’application point-final dans l’approximation nilpotente de la variété.Nous appliquons des méthodes perturbatives pour obtenir des asymptotiques sur la longueur de courbes localement minimisantes perdant leur optimalité proche de leur point de départ dans le cas des variétés sous-riemanniennes de contact de dimension arbitraire. Nous décrivons la géométrie du lieu singulier et prouvons sa stabilité dans le cas des variétés de dimension 5.Nous introduisons une construction permettant de définir des champs de directions à l’aide de couples de champs de vecteurs. Ceci fournit une topologie naturelle pour analyser la stabilité des singularités de champs de directions sur des surfaces. / We investigate the relationship between features of of sub-Riemannian geometry and an array of singularities that typically arise in this context.With sub-Riemannian Whitney theorems, we ensure the existence of global extensions of horizontal curves defined on closed set by requiring a non-singularity hypothesis on the endpoint-map of the nilpotent approximation of the manifold to be satisfied.We apply perturbative methods to obtain asymptotics on the length of short locally-length-minimizing curves losing optimality in contact sub-Riemannian manifolds of arbitrary dimension. We describe the geometry of the singular set and prove its stability in the case of manifolds of dimension 5.We propose a construction to define line fields using pairs of vector fields. This provides a natural topology to study the stability of singularities of line fields on surfaces.

No description available.

Internal consistency reliability and validity were established for a new 31 item Imagery Manipulation Scale. Previous attempts to correlate subjectively rated control of visual imagery with tests of spatial ability have been unsuccessful. However, no attempt to construct a subjectively rated control of imagery scale was located which tried to establish internal consistency reliability and both content and construct validity. Further, no research was located in which subjects were requested to rate their imagery ability utilized during the performance of the actual spatial tasks used to try to establish validity. A new scale of subjectively rated control of imagery was devised in which subjects were requested to rate their imagery while solving spatial tasks which involved visualizing the manipulation of geometric forms. Content validity was established by analyzing the transformation involved while solving the spatial problems. Internal consistency reliability for the 31 item scale was established across two samples. Validity was established with the second sample (100 university students: 26 male and 74 female). The task utilized to provide validity could be objectively scored, and was made up of four spatial subtests, which were adapted from the Vandenberg and Kuse Mental Rotations Test, the Kosslyn Directions Test, performed in both the forward and backward direction, and a block task utilized by Snyder. A convergent and discriminant validity analysis established construct validity. Further, the hypotheses of three investigators, Kosslyn, Shepard and his colleagues, and Snyder, were supported by the results of the present investigation, thus substantiating the conclusion that reported control of imagery processing can be operationalized with performance scores on spatial ability tasks.

In order to better identify and understand the differences between Japanese and English, the task of giving walking directions was used. Japanese and American public facilities (10 each) were randomly chosen from which to collect data over the phone in order to examine these differences based on the following five communication styles: 1) politeness, 2) indirectness, 3) self-effacement, 4) back-channel feedback (Aizuchi), 5) and other linguistic and cognitive differences in relation to space and giving directions. The study confirmed what the author reviewed in the literature: Japanese are more polite, English speakers tend to give directions simply and precisely, Japanese prefer pictorial information and most Americans prefer linguistic information, Japanese is a topic-oriented language and also an addressee-oriented language. The information revealed from this study will help Japanese learners develop important skills needed for developing proficiency in the target language and also teach important differences between the two languages.

This study is about three French plays in which Jocasta, the mother and wife of Oedipus, is shared as a main character: La Thébaïde (The Theban Brothers) by Jean Racine, La Machine Infernale (The Infernal Machine) by Jean Cocteau, and Le Nom d’Oedipe (The Name of Oedipus) by Hélène Cixous. Jocasta has always been overshadowed by the tragic destiny of Oedipus since the onset of Sophocles’ works. Although these three plays commonly focus on describing the character of Jocasta, there are some remarkable differences among them in terms of theme, style, and stage directions. In The Theban Brothers, Racine’s 17th century play, Jocasta is described as a deathlike mother, while Cocteau’s Jocasta, in The Infernal Machine, is portrayed as an “extravagant, liberal, and hilarious” lady. In The Name of Oedipus, Cixous portrays Jocasta as a woman possessing hermaphroditic characteristics, ushering in a new era of resistance to the age-old paternal hierarchy. As for style, Racine’s neoclassical play shows a strict respect for the three unities of time, space, and action. Cocteau’s avant-garde play neglects all these rules, while Cixous goes even further by destroying the order of languages, as illustrated by her “feminine writing.” Freed from Western orthodoxy, Cixous wants to contribute to the creation of cosmic unity. Her deconstructionist play intends to regenerate the world by establishing a new order and new point of view towards universality. The stage directions of these plays are also an important key to better understanding theatrical evolution. It is through the stage directions, indicated both implicitly and explicitly in these three plays, iv that enables us to appreciate the theatrical transformation in terms of visualization as well as metaphysics. In sum, the transformation of theme, style, and stage devices in portraying their own Jocastas demonstrates that while these three plays are deconstructional to one another, each denying the existing value and orders of their respective time periods, they are also constructional in that they all attempt to open a new horizon of theatre

Optical testing in adverse environments, ophthalmology and applications where characterization by curvature is leveraged all have a common goal: accurately estimate wavefront shape. This dissertation investigates wavefront sensing techniques as applied to optical testing based on gradient and curvature measurements. Wavefront sensing involves the ability to accurately estimate shape over any aperture geometry, which requires establishing a sampling grid and estimation scheme, quantifying estimation errors caused by measurement noise propagation, and designing an instrument with sufficient accuracy and sensitivity for the application. Starting with gradient-based wavefront sensing, a zonal least-squares wavefront estimation algorithm for any irregular pupil shape and size is presented, for which the normal matrix equation sets share a pre-defined matrix. A Gerchberg–Saxton iterative method is employed to reduce the deviation errors in the estimated wavefront caused by the pre-defined matrix across discontinuous boundary. The results show that the RMS deviation error of the estimated wavefront from the original wavefront can be less than λ/130~ λ/150 (for λ equals 632.8nm) after about twelve iterations and less than λ/100 after as few as four iterations. The presented approach to handling irregular pupil shapes applies equally well to wavefront estimation from curvature data. A defining characteristic for a wavefront estimation algorithm is its error propagation behavior. The error propagation coefficient can be formulated as a function of the eigenvalues of the wavefront estimation-related matrices, and such functions are established for each of the basic estimation geometries (i.e. Fried, Hudgin and Southwell) with a serial numbering scheme, where a square sampling grid array is sequentially indexed row by row. The results show that with the wavefront piston-value fixed, the odd-number grid sizes yield lower error propagation than the even-number grid sizes for all geometries. The Fried geometry either allows sub-sized wavefront estimations within the testing domain or yields a two-rank deficient estimation matrix over the full aperture; but the latter usually suffers from high error propagation and the waffle mode problem. Hudgin geometry offers an error propagator between those of the Southwell and the Fried geometries. For both wavefront gradient-based and wavefront difference-based estimations, the Southwell geometry is shown to offer the lowest error propagation with the minimum-norm least-squares solution. Noll’s theoretical result, which was extensively used as a reference in the previous literature for error propagation estimate, corresponds to the Southwell geometry with an odd-number grid size. For curvature-based wavefront sensing, a concept for a differential Shack-Hartmann (DSH) curvature sensor is proposed. This curvature sensor is derived from the basic Shack-Hartmann sensor with the collimated beam split into three output channels, along each of which a lenslet array is located. Three Hartmann grid arrays are generated by three lenslet arrays. Two of the lenslets shear in two perpendicular directions relative to the third one. By quantitatively comparing the Shack-Hartmann grid coordinates of the three channels, the differentials of the wavefront slope at each Shack-Hartmann grid point can be obtained, so the Laplacian curvatures and twist terms will be available. The acquisition of the twist terms using a Hartmann-based sensor allows us to uniquely determine the principal curvatures and directions more accurately than prior methods. Measurement of local curvatures as opposed to slopes is unique because curvature is intrinsic to the wavefront under test, and it is an absolute as opposed to a relative measurement. A zonal least-squares-based wavefront estimation algorithm was developed to estimate the wavefront shape from the Laplacian curvature data, and validated. An implementation of the DSH curvature sensor is proposed and an experimental system for this implementation was initiated. The DSH curvature sensor shares the important features of both the Shack-Hartmann slope sensor and Roddier’s curvature sensor. It is a two-dimensional parallel curvature sensor. Because it is a curvature sensor, it provides absolute measurements which are thus insensitive to vibrations, tip/tilts, and whole body movements. Because it is a two-dimensional sensor, it does not suffer from other sources of errors, such as scanning noise. Combined with sufficient sampling and a zonal wavefront estimation algorithm, both low and mid frequencies of the wavefront may be recovered. Notice that the DSH curvature sensor operates at the pupil of the system under test, therefore the difficulty associated with operation close to the caustic zone is avoided. Finally, the DSH-curvature-sensor-based wavefront estimation does not suffer from the 2π-ambiguity problem, so potentially both small and large aberrations may be measured.

Актуальность диссертационной работы заключается в необходимости изучения взаимосвязи между технологиями коммуникации среди поколений в семье и снижением социальных девиаций в молодежной среде. Изучены теоретические основания исследования межпоколенческих коммуникаций и социальных девиаций. Разработаны технологии межпоколенческой коммуникации, предложена их классификация по трем основаниям и описаны три актуальные технологии. В ходе социологического опроса выявлено наличие проблем во взаимоотношениях между поколениями в семье, связанных с непониманием друг друга. Для решения выявленных проблем разработан проект, в рамках которого укрепляются взаимоотношения между поколениями в семье, проводятся междисциплинарные мероприятия для специалистов сферы превентологии. / The relevance of the dissertation work lies in the need to study the relationship between communication technologies among generations in the family and the reduction of social deviations in the youth environment. The theoretical foundations of the study of intergenerational communications and social deviations are studied. Technologies of intergenerational communication have been developed, their classification on three grounds has been proposed, and three current technologies have been described. During the sociological survey, the presence of problems in the relationship between generations in the family associated with misunderstanding of each other was revealed. To solve the identified problems, a project has been developed within the framework of which relationships between generations in the family are strengthened, interdisciplinary events are held for specialists in the field of preventology.

[pt] O contato com modelos geométricos tais como planos, esferas ou quádricas é uma importante ferramenta para entender a geometria diferencial de uma superfície. Nesta tese, estudamos o contato de superfícies com quádricas, e mais particularmente, com as quádricas de Moutard. Estendemos os resultados conhecidos para o caso de pontos parabólicos e curvas flecnodais em superfícies genéricas. Consideramos também o contato de hipersuperfícies com hiperquádricas de Moutard. / [en] The contact with generic models like planes, spheres or quadrics is an important tool to understand the differential geometry of a surface. In this thesis, we study the contact of surfaces with quadrics, more specifically, with Moutard quadrics. We extend the known results for the case of parabolic points and flecnodal curves in generic surfaces. We consider also the contact of hypersurfaces with Moutard hyperquadrics.