Thesis (M. Th.)--Capital Bible Seminary, 1973. / Includes bibliographical references (leaves 68-72).

Proefschrift--Leiden--Rijksuniversiteit, 1993. / Résumé en anglais. Bibliogr. p. 359-365. Index.

Diss.--Wirtschaftswissenschaftlichen Fakultät$--Münster--Westfälichen Wilhelms-Universität, 1993. / Bibliogr. p. 261-279.

Th. 3e cycle--Math. pures--Lille 1, 1979. N°: 775.

As of 2013, 39,000 Chileans have been diagnosed with HIV/AIDS; HIV (Human Immunodeficiency Virus) has a prevalence of 8.6 cases per hundred thousand people, and AIDS (Acquired Immune Deficiency Syndrome) has a prevalence of 5.4 cases per hundred thousand people. In addition, 99.2% of the cases have been reported as having been sexually transmitted. The objective of this study was to determine if a relationship existed between education level and HIV preventive actions. HIV preventive actions include the access to condoms, HIV-related diagnostic exams, and the dissemination of accurate preventative information. A descriptive analysis with data from the NIH-funded project, "Testing an HIV/AIDS Prevention Intervention for Chilean Women," was conducted. A total of four hundred women were included in this analysis (n=400). Pearson correlation was conducted to determine the strength of the relationships between the highest level of education completed and Papanicolau examinations, condom purchases or usage, sexual activity, communication with sexual partner and HIV/AIDS testing. We expect to see a strong positive correlation between education level vi and HIV-preventive actions-the lower the education level obtained, the less likely participants are to engage in HIV-preventive actions. The results of this study failed to demonstrate a correlation between education level and health-related preventive actions as demonstrated by the parameters evaluated here. The level of education of the participants did not correlate with STI diagnoses (r=- 0.16, p=0.747), number of sexual partners in her lifetime (r= -0.003, p=0.954), number of sexual partners in the previous three months (r= -0.016, p=0.751), communication with sexual partner regarding condom usage (r= -0.033, p=0.516), condom purchases(r=0.001, p=0.987), Papanicolau examination (r=-0.015, p=0.789), nor did it correlate with HIV testing rates (r= -0.005, p=0.92). No distinction was made regarding type of institution participants attended. From the results obtained, four main areas were identified to affect health-related preventive actions amongst the Chilean women interviewed including (1) Participants reported under-utilization of diagnostic exams, (2) Decreased use of condoms, (3) Cultural stigma towards HIV/AIDS and (4) Lack of continuity of care in women's health The next step for Chilean Public Health officials is to create public health programs and campaigns to increase population knowledge and understanding of HIV and AIDS. Secondary education must integrate information regarding safe sex practices into the curriculum. Future studies should investigate type of schooling attended (government-subsidized, semi-private, private), to determine variations in quality of education between socioeconomic status and the association with health-related preventive actions.

Cette thèse traite de la reconnaissance automatique d'action humaine dans des vidéos. La reconnaissance d'action humaine est indispensable pour déterminer quelles actions humaines se produisent dans des vidéos. Ce problème est particulièrement difficile en raison d'énormes variations dans les aspects visuels et de mouvement des personnes et des actions, les changements de point de vue de la caméra, le fond mobile, des occlusions, la présence de bruit, ainsi que l'énorme quantité de données vidéos. Tout d'abord, nous passons en revue, évaluons et comparons les techniques les plus importantes et les plus populaires de l'état de l'art pour la reconnaissance d'action, ensuite, nous proposons une plateforme basée sur des caractéristiques locales, que nous utilisons tout au long de ce travail de thèse pour étudier de nouveaux algorithmes. En plus, nous introduisons une nouvelle base de données (Hôpital CHU de Nice) avec des actions de la vie quotidienne de patients âgés dans cet hôpital. Ensuite, nous proposons deux descripteurs spatio-temporels locaux pour la reconnaissance d'action dans les vidéos. Le premier descripteur est basé sur une représentation des matrices de covariance, modélisant les relations linéaires entre les caractéristiques bas niveaux. Le deuxième descripteur est basé sur les covariances browniennes, et modélise tous les types de relations possibles entre les caractéristiques bas niveaux. Après, nous proposons trois représentations de caractéristiques de hauts niveaux pour dépasser les limites des techniques utilisant l'encodage des sacs de mots. La première représentation est basée sur le principe des trajectoires relatives denses. Nous proposons une représentation objet-centrée des caractéristiques locales des trajectoires de mouvement, ce qui permet d'utiliser l'information spatiale par une technique de codage des caractéristiques locales. La deuxième représentation encode les relations entre les caractéristiques locales par paires. Le principe est dextraire les relations d'apparence entre les caractéristiques (à la fois visuelles et de mouvement), et dutiliser l'information géométrique pour décrire la façon dont ces relations d'apparence sont disposées mutuellement dans l'espace spatio-temporel. La troisième représentation calcule les statistiques des paires concomitantes des mots visuels dans les voisinages multi-échelles centrées les caractéristiques. La représentation basée sur les caractéristiques contextuelles proposées encode linformation sur la densité locale de ces caractéristiques, les relations entre les paires des caractéristiques locales et leur ordre spatio-temporel. Finalement, les techniques proposées permettent d'obtenir une performance meilleure ou semblable par rapport à l'état de l'art, sur des bases de données représentant une grande diversité dactions humaines (Weizmann, KTH, URADL, MSR Daily Activity 3D, HMDB51, et Hôpital CHU de Nice). / This thesis targets the automatic recognition of human actions in videos. Human action recognition is defined as a requirement to determine what human actions occur in videos. This problem is particularly hard due to enormous variations in visual and motion appearance of people and actions, camera viewpoint changes, moving background, occlusions, noise, and enormous amount of video data. Firstly, we review, evaluate, and compare the most popular and the most prominent state-of-the-art techniques, and we propose our action recognition framework based on local features, which we use throughout this thesis work embedding the novel algorithms. Moreover, we introduce a new dataset (CHU Nice Hospital) with daily self care actions of elder patients in a hospital. Then, we propose two local spatio-temporal descriptors for action recognition in videos. The first descriptor is based on a covariance matrix representation, and it models linear relations between low-level features. The second descriptor is based on a Brownian covariance, and it models all kinds of possible relations between low-level features. Then, we propose three higher-level feature representations to go beyond the limitations of the local feature encoding techniques. The first representation is based on the idea of relative dense trajectories. We propose an object-centric local feature representation of motion trajectories, which allows to use the spatial information by a local feature encoding technique. The second representation encodes relations among local features as pairwise features. The main idea is to capture the appearance relations among features (both visual and motion), and use geometric information to describe how these appearance relations are mutually arranged in the spatio-temporal space. The third representation captures statistics of pairwise co-occurring visual words within multi-scale feature-centric neighbourhoods. The proposed contextual features based representation encodes information about local density of features, local pairwise relations among the features, and spatio-temporal order among features. Finally, we show that the proposed techniques obtain better or similar performance in comparison to the state-of-the-art on various, real, and challenging human action recognition datasets (Weizmann, KTH, URADL, MSR Daily Activity 3D, HMDB51, and CHU Nice Hospital).

Notre thèse, qui s’inscrit dans le champ de la linguistique interactionnelle. Elle porte sur les interactions au cours du repas entre amis dans les invitations en France. Basée sur les données audio et vidéo enregistrées en situation naturelle, cette étude empirique se donne pour but de montrer les modalités par lesquelles les participants co-construisent les deux activités principales déployées à table : « manger » et « parler ». Concernant l’activité de « manger », notre travail décrit en détail les interactions à partir du moment où les mangeurs s’installent à table jusqu’au moment où ils finissent le repas. Grâce à l’analyse verbale et multimodale, notre étude dégage les ressources linguistiques et multimodales mises en œuvre par les hôtes et les invités pour assurer le bon déroulement du repas. Quant à l’activité de « parler », elle est abordée à travers l’évaluation des plats servis. Notre analyse montre qu’à travers cette activité, les locuteurs expriment leur appréciation personnelle vis-à-vis des mets, en même temps qu’ils réalisent différents actions telles que complimenter, critiquer, s’auto-complimenter et s’auto-déprécier. L’enjeu de cette thèse est de mettre en lumière les principes connus mais inaperçus par lesquels les locuteurs français construisent l’interaction à table. / My dissertation, which lies within the framework of international linguistics, is about the interaction during meals among friends in invitations in France. Based on audio and video data recorded in naturally-occurring situations, this empirical study aims to show the modalities by which the participants co-construct the two main activities observed at table: eating and talking. Concerning eating, the study describes in detail the interactions since the eaters sit down at the table until they finish the meal. On the basis of verbal and multimodal analysis, this study brings out the linguistic and multimodal resources used by the hosts and the guests in order to make sure of the course of the meal. As for talking, this dissertation focuses on the assessment on the dishes served. The analysis shows that through the assessment on the dishes served, the speakers express their personal observation about the food, as well, they realize various actions such as complimenting, criticizing, self-complimenting and self-depreciating. The key issue of this study is to shed light on the principles seen but unnoticed by which the French speakers construct interaction at table.

No description available.

In this thesis, I analyse apologies produced by British political figures from a pragmatic perspective. In particular, I seek to explain the function of political apologies and describe the form they take. In order to give a thorough account of the speech act of apologising in the public sphere, I look to a variety of genres for data. The set of remedial acts scrutinised in this study come from debates and statements in the House of Commons, the Leveson Inquiry and news interviews. The differences in communicative practices between these data sources mean that the types of apology that come about within each genre are varied. Many of the parliamentary apologies are monologic, whereas the apologetic actions found at the Leveson Inquiry and in news interviews are dialogic and, to some extent, co-constructed between participants. These differences mean that a variety of theoretical approaches are taken in analysing the data – speech act theory (Austin, 1962; Searle, 1969) and generalised conversational implicature theory (Levinson,2000) feature heavily in the discussion of monologic apologies. Apologies produced within an interactive, ‘conversational’ setting are treated using developments in conversation analysis (amongst others see: Sacks, 1992; Schegloff, 2007). I attempt to reconcile these two, quite different, approaches to discourse at various points in the thesis, arguing that conversation analysis lacks a theory of how interlocutors understand what actions are happening in interaction (and this is provided by speech act theory) and speech act theory lacks a detailed focus on what actually happens in language as interaction (provided by conversation analysis). On the basis of the apology data scrutinised in the thesis, I propose a set of felicity conditions for the speech act of apology (chapter 2) and discuss how the apology (and speech acts broadly) should be considered as prototype entities (chapter 8). I show that when apologising for actions which they have committed, politicians are more fulsome in their apologies than we are in everyday conversation. I also show that they use more explicit apology tokens than is found in quotidian talk (chapter 3). When apologising for historical wrongs, I demonstrate that apologising is a backgrounded act and the focus of the statement is on being clear and unequivocal about the nature of the offences for which the government is apologising (chapter 6). I also argue that political apologies in interactive settings are best thought of as action chains (Pomerantz, 1978). That is to say, apologies in these environments may elicit a response from an interlocutor, but do not need to (chapters 4 & 5). This is quite unlike everyday talk (cf Robinson, 2004). I discuss how apology tokens may be used in the performance of other acts, including introducing dissent and undertaking serious face threat. I suggest that this comes about because apology tokens exist on a cline of pragmaticalisation (chapter 7).

Let X be a smooth projective variety over C equipped with a C*-action whose fixed points are isolated. Let Y and Z be subvarieties of complementary dimentions in X which intersect properly. In this thesis we present an algorithm for computing the points of intersection between Y and Z based on homotopy continuation and the Bialynicki-Birula decompositions of X into locally closed invariant subsets. As an application we present a new solution to the inverse kinematic problem of a general six-revolute serial-link manipulator. / QC 20101108