From a literary review of the so-called continuous past tenses of the Bantu languages with special reference to Zulu it becomes quite clear that a number of misconceptions prevail amongst grammarians concerning these tense forms. Traditional grammarians refer to these verbal forms using labels such as continuous past tenses, compound tenses, the perfect form of the past tense, past compound tenses, contingent form/mood/tense, participial tenses, compound tenses and moods and imperfective aspect. These terms are all inappropriate names for these tenses. Posthumus (1983, 1988, 1990, 1999) is the only Bantuist who refers to the relative tense forms of Zulu as the relative tenses. His analysis and description of the grammatical tenses of Zulu are based on the Reichenbach model for tense analysis. Even though a number of European grammarians (initially especially German researchers) followed the Reichenbach model, none of the Bantuists (except Posthmus) employed this model of tense analysis2. In fact, the Bantuists. analyses of tense (as contained in grammars, hand books and articles in scientific journals) are not scientifically grounded. The distinction that Posthumus draws between absolute and relative tenses is also adhered to in this research. It is essential to distinguish clearly between tense and time reference on the one hand and between tense and aspect on the other hand because grammarians frequently confuse these categories. A clear distinction must, furthermore, be made between absolute and relative tense. Posthumus supplied a scientifically sound description of the structural and semantic characteristics of the relative tenses and has included a schematic representation of the array of relative tenses that occur in Zulu - however, three aspects of these tenses remain that have not been addressed adequately in his publications that will be scrutinized in this research. Firstly, the full spectrum of relative tenses with .be will be given and a scientifically sound analysis will be given thereof. A relative tense is defined as the verb form that marks the relationship between coding time, reference time and event time. Secondly, the contracted forms of the relative tenses are recorded and analysed systematically. The derivation of the abridged forms from the original full auxiliary verb groups is indicated and the various variants are listed in table form. Thirdly, attention is devoted to finding an appropriate label for the so-called compound tenses (as a super ordinate term) and to finding suitable names for the individual relative tense forms. The point of departure in this study is that the appropriate naming of scientific grammatical categories is of the utmost importance because that prevents misunderstanding. In fact, it is partly because of the use of inappropriate terms for the relative tenses collectively, and for the individual relative tense forms, that consecutive generations of Bantu grammarians have continued to offer inapt descriptions of these language forms. This state of affairs is quite evident from the literary survey. A thorough literary survey of the traditional descriptions and the labelling of the relative tenses are also aspects that have not received adequate attention in Posthumus.s publications and will therefore be pursued in this research. From the literary survey of the relative tenses it transpires that different and sometimes opposing views exist concerning their description and that different names exist for the so-called compound tenses (most of them being totally inapplicable). A comprehensive literary survey focussing on the description of the so-called compound past tenses is essential to put the traditional descriptions and analyses of these tenses in perspective. Such a literary survey was undertaken and included all the official Bantu languages of South Africa except South Ndebele. Southern Ndebele was not included in the survey because there is no grammar available on this language and information on the relative tenses of this language is not readily available either. An outline distinction is drawn in this study between the categories tense, time reference and aspect (which are often confused by grammarians). A clear distinction is also drawn between absolute and relative tenses. An exhaustive description of the structural characteristics of the individual relative tenses is offered. In this discussion the relative tense forms consisting of an auxiliary verb group (comprising the auxiliary verb stem .be diachronically derived from the copulative verb stem .ba) are discussed in detail. The structural and semantic characteristics of both the auxiliary verb part and the complementary part are discussed in detail in this study. An exposition is supplied of the complement that may be a main verb (as for instance in the example zibe zigijima (> bezigijima) izingane ... .they were busy running, the children ....), an auxiliary verb group (as in the example zibe zinele zigijime (> bezinele zigijime) izingane ... .as soon as they are busy running, the children ....) or a copulative word group (as in the example zibe ziba zinkulu (> beziba zinkulu) izingane ... .they were busy becoming big, the children ... / they were becoming grown up, the children ....). However, it is not possible to discuss the different types of complements exhaustively in this research report, therefore the focus will be on those examples where a main verb occurs as complement. The semantic traits of the relative tenses are discussed and finally names are proposed as labels for the individual relative tenses that occur in Zulu. The names proposed for the individual relative tenses are descriptive terms based on the semantic characteristics of these tense forms. The terms suggested for the sixteen individual relative tenses are motivated and listed in a systematic exposition. Due to the limited nature of a research report of this nature, the naming of the relative tense forms with a copulative as complement are not discussed. Furthermore, mere mention is made of the possible inclusion of the (deficient) auxiliary verb stem .se in the relative tense forms. A comprehensive discussion of this aspectual distinction also falls outside the scope this mini-dissertation. / Dissertation (MA)--University of Pretoria, 2008. / African Languages / unrestricted

Les effets indésirables liés aux médicaments (EIM) sont des dommages liés au traitement médicamenteux plutôt qu’aux conditions sous-jacentes du patient. Ils mettent les patients en danger, et la plupart d’entre eux sont évitables. La détection des EIM repose habituellement sur les reports spontanés d’EIM et sur la revue de dossiers. L’objectif du présent travail est d’identifier automatiquement les cas d’EIM en utilisant des méthodes de Data Mining (fouille statistique de données). Le DataMining est un ensemble de méthodes statistiques particulièrement adaptées à la découverte de règles dans de grandes bases de données.Matériel Un modèle de données commun est tout d’abord défini, dans le but de décrire les données qui peuvent être extraites des dossiers patient électroniques. Plus de 90 000séjours hospitaliers complets sont extraits de 5 hôpitaux français et danois. Ces enregistrements incluent les diagnostics, les résultats de biologie, les médicaments administrés, des informations démographiques et administratives, et enfin du texte libre (courriers, comptes-rendus). Lorsque les médicaments ne peuvent être extraits d’un CPOE (système de prescription connectée), ils sont extraits des courriers pa rSemantic Mining (fouille de texte). De plus, la société Vidal fournit un ensemble exhaustif de RCP (Résumés des Caractéristiques du Produit).Méthode On tente de tracer dans les données tous les événements indésirables décrits dans les RCP. Puis en utilisant les méthodes de Data Mining, en particulier les arbres de décision et les règles d’association, on identifie les circonstances qui favorisent l’apparition d’EIM. Plusieurs règles de détection des EIM sont ainsi obtenues, elles sont ensuite filtrées et validées par un comité d’experts. Enfin, les règles sont décrites sous forme de fichiers XML et stockées dans une base. Elles sont exécutées afin de calculer certaines statistiques et de détecter les cas d’EIM.Résultats236 règles de détection des EIM sont ainsi découvertes. Elles permettent de détecter27 types d’événements indésirables différents. Plusieurs statistiques sont calculées automatiquement pour chaque règle dans chaque service, comme la confiance ou le risque relatif. Ces règles impliquent des conditions innovantes : par exemple certaines règles décrivent les conséquences de l’arrêt d’un médicament. De plus, deux outils Web sont développés et mis à la disposition des praticiens via Internet : les Scorecards permettent de présenter des informations statistiques e tépidémiologiques sur les EIM propres à chaque service, tandis que l’Expert Explorer permet aux médecins d’examiner en détail les cas probables d’EIM de leur service.Enfin, une évaluation préliminaire de l’impact clinique des EIM est menée, ainsi que l’évaluation de la précision de détection des EIM. / Adverse Drug Events (ADE) are injuries due to medication management rather than the underlying condition of the patient. They endanger the patients and most of them could be avoided. The detection of ADEs usually relies on spontaneous reporting ormedical chart reviews. The objective of the present work is to automatically detectcases of ADEs by means of Data Mining, which are a set of statistical methodsparticularly suitable for the discovery of rules in large datasets.MaterialA common data model is first defined to describe the available data extracted from the EHRs (electronic health records). More than 90,000 complete hospital stays areextracted from 5 French and Danish hospitals. Those complete records includediagnoses, lab results, drug administrations, administrative and demographic data aswell as free-text reports. When the drugs are not available from any CPOE(Computerized Prescription Order Entry), they are extracted from the free-text reports by means of semantic mining. In addition, an exhaustive set of SPCs (Summaries of Product Characteristics) is provided by the Vidal Company.MethodsWe attempt to trace all the outcomes that are described in the SPCs in the dataset. By means of data mining, especially Decision Trees and Association Rules, the patternsof conditions that participate in the occurrence of ADEs are identified. Many ADEdetection rules are generated; they are filtered and validated by an expert committee. Finally, the rules are described by means of XML files in a central rules repository, and are executed again for statistics computation and ADE detection.Results236 ADE-detection rules have been discovered. Those rules enable to detect 27different kinds of outcomes. Several statistics are automatically computed for eachrule in every medical department, such as the confidence or the relative risk. Thoserules involve innovative conditions: for instance some of them describe theconsequences of drug discontinuations.In addition, two web tools are designed and are available through the web for thephysicians of the departments: the Scorecards enable to display statistical andepidemiological information about ADEs in a given department and the ExpertExplorer enables the physicians to review the potential ADE cases of theirdepartment.Finally, a preliminary evaluation of the clinical impact of the potential ADEs isperformed as well as a preliminary evaluation of the accuracy of the ADE detection.

Psychologists interested in the study of language find that people are faster at making decisions about words that are related than they are at making decisions about words that are not related – an effect called semantic priming. This phenomenon has largely only been document in laboratory settings using natural languages as contest and real words as stimuli. The current study explores the relation between the semantic priming effect and a laboratory procedure designed to give rise to performances that can be described as linguistic. Six adult participants learned to partition a collection of eight stimuli into two sets of four stimuli. Following this, the subjects showed the semantic priming effect within a set of stimuli but not across sets. These data suggest that it may be possible to study linguistic phenomenon in laboratory-based procedures allowing better control and the ability to ask very precise questions about linguistic functioning.

In parallel passages from Old and Middle English and in noun cognates from Modern English, Old English, and Modern German, the most discernible elements of language drift are gender loss and word meaning change, respectively. They can be observed, discussed, and calculated to show a definite progression toward the development of Modern English.

Orientador: Giovanna Longo / Resumo: Este trabalho é composto por pesquisa bibliográfica voltada para o ensino lexical nas aulas de língua portuguesa. Através dos estudos realizados, podemos perceber que a questão lexical, apesar da sua importância, não é considerada um tema relevante nas salas de aula, tornando-se marginalizada no ensino. Há tempos, o estudo voltado para a gramática tem ficado em primeiro lugar nas aulas de língua materna, pois de acordo com os conceitos culturais impostos por nossa sociedade, só se fala bem o idioma se se conhece impecavelmente a gramática. No aprofundamento das leituras e no acompanhamento pedagógico, nota-se essa errônea ideia. Havendo uma imensa gama de variedades linguísticas e possibilidades semânticas, conhecer a gramática, nada mais é que conhecer a norma padrão instituída pela escrita, ignorando a pluralidade e mutabilidade do idioma como patrimônio cultural. Percebe-se, portanto, a necessidade urgente de mudança nas nossas práticas didáticas, introduzindo textos e dicionários, focando o estudo lexical, mas não de uma forma irrelevante e pouco significativa, mas sim de maneira a levar o discente a perceber sua língua como algo vivo, em constante modificação, tendo a habilidade de buscar no acervo lexical a melhor maneira de representar-se no mundo e poder ver-se como sujeito deste mundo / Abstract: This paper is comprised of a bibliographical study focusing on lexical teachings for the Portuguese language as a school subject. Through this research it was possible to ascertain that the lexical aspect of the Portuguese language is not considered a relevant topic in classes, despite its importance, marginalizing it in the education. For a long time now, grammar has had the utmost importance when teaching Portuguese because of cultural beliefs: in order to speak proper Portuguese you need to flawlessly know its grammar. With further reading on the subject as well as pedagogic supervision this is evidenced to be untrue. Having a broad array of linguistics variants and semantic possibilities, knowing grammar is - essentially - knowing the standard norm established by written Portuguese while ignoring the pluralism and mutability of the Portuguese language, its cultural heritage. Therefore, it can be seen that there is a necessity for changes in the teaching practices by introducing texts and dictionaries focusing on lexical studies in a relevant and significant manner so as to have the student perceive the Portuguese language as being alive and in constant change, giving them the ability to seek in its lexical collection the best way to present themselves and feel like they belong. / Mestre

<p>Despite the rapid development of adversarial attacks on machine learning models, many types of new adversarial examples remain unknown. Undiscovered types of adversarial attacks pose a</p><p>serious concern for the safety of the models, which raises the issue about the effectiveness of current adversarial robustness evaluation. Image semantic segmentation is a practical computer</p><p>vision task. However, segmentation networks’ robustness under adversarial attacks receives insufficient attention. Recently, machine learning researchers started to focus on generating</p><p>adversarial examples beyond the norm-bound restriction for segmentation neural networks. In this thesis, a simple and efficient method: AdvDRIT is proposed to synthesize unconstrained controllable adversarial images leveraging conditional-GAN. Simple CGAN yields poor image quality and low attack effectiveness. Instead, the DRIT (Disentangled Representation Image Translation) structure is leveraged with a well-designed loss function, which can generate valid adversarial images in one step. AdvDRIT is evaluated on two large image datasets: ADE20K and Cityscapes. Experiment results show that AdvDRIT can improve the quality of adversarial examples by decreasing the FID score down to 40% compared to state-of-the-art generative models such as Pix2Pix, and also improve the attack success rate 38% compared to other adversarial attack methods including PGD.</p>

Image interpolation in the medical area is of vital importance as most 3D biomedical volume images are sampled where the distance between consecutive slices is significantly greater than the in-plane pixel size due to radiation dose or scanning time. Image interpolation creates a certain number of new slices between known slices in order to obtain an isotropic volume image. The results can be used for the higher quality of 2D and 3D visualization or reconstruction of human body structure. Semantic interpolation on the manifold has been proved to be very useful for smoothing the interpolation process. Nevertheless, all previous methods focused on low-resolution image interpolation, and most of which work poorly on high-resolution images. Besides, the medical field puts a high threshold for the quality of interpolations, as they need to be semantic and realistic enough, and resemble real data with only small errors permitted. Typically, people downsample the images into 322 and 642 for semantic interpolation, which does not meet the requirement for high-resolution in the medical field. Thus, we explore a novel way to generate semantically correct interpolations and maintain the resolution at the same time. Our method has been proved to generate realistic and high-resolution interpolations on the sizes of 5262 and 5122. Our main contribution is, first, we propose a novel network, High Resolution Interpolation Network (HRINet), aiming at producing semantically correct high-resolution CT image interpolations. Second, by combining the idea of ACAI and GANs, we propose a unique alternative supervision method by applying supervised and unsupervised training alternatively to raise the accuracy and fidelity of body structure in CT when interpolated while keeping high quality. Third, we introduce an extra Markovian discriminator as a texture or fine details regularizer to make our model generate results indistinguishable from real data. In addition, we explore other possibilities or tricks to further improve the performance of our model, including low-level feature maps mixing, and removing batch normalization layers within the autoencoder. Moreover, we compare the impacts of MSE based and perceptual based loss optimizing methods for high quality interpolation, and show the trade-off between the structural correctness and sharpness. The interpolation experiments show significant improvement on both sizes of 256 2 and 5122 images quantitatively and qualitatively. We find that interpolations produced by HRINet are sharper and more realistic compared with other existing methods such as AE and ACAI in terms of various metrics. As an application of high-resolution interpolation, we have done 2D volume projection and 3D volume reconstruction from axial view CT data and their interpolations. We show the great enhancement of applying HRINet for both in sharpness and fidelity. Specifically, for 2D volume projection, we explore orthogonal projection and weighted projection respectively so as to show the improved effectiveness for visualizing internal and external human body structure.

The purpose of this investigation was to assess the validity of utilizing the same semantic differential test as a measure of social evaluation for persons of different ages. The semantic differential, a rating technique by which judgments of stimuli are made on seven-interval bipolar descriptive continua called "scales", currently receives widespread usage as a measure of attitudes toward aging and the elderly. However, a lack of semantic stability across stimuli and across subjects, known, respectively, as "concept-scale" and "subject2 scale" interaction, has been found to occur with the technique in various areas of research. That such a lack of stability might occur across stimuli and/or subjects of different ages is suggested by the existence of . di fferences between people of different ages which affect both the appearance and functioning of the individual. These differences derive from the biological aging process, the life cycle process, and the social change process. This study investigated whether such cross-age differences are of sufficient severity to cause people of different ages to be perceived as different classes of stimuli and/or to perceive others as different populations of subjects. Ratings of eight videotaped stimulus models were made by 60 younger (aged 22 to 32) and 60 older (aged 60 and older ) volunteer subjects on a semantic differential test composed of 38 scales. The stimulus models, consisting of four younger and four older adults, were non-actors, unknown to the rating subjects, and presented for one minute in a standardized visual format and without sound. Of the 38 scales used, seven were selected as reference scales from earlier studies, 30 were suggested by a volunteer "generating" sample of 30 younger and JO older subjects, and one (young/old) was included as a check on the perceived ages of the stimulus models. Scale scores were factor analyzed to establish the underlying factor structure of social evaluation and to ascertain whether it rema1ned stable across stimulus model and subject age levels. Four separate factor analyses were performed -- younger stimulus models/younger subjects, younger stimulus models/older subjects, older stimulus J models/older subjects, and older stimulus models/younger subjects -- permitting comparisons between the factor matrices for concept-scale and subject-scale interaction. Three factors were defined by each of the four separate factor analyses of the scales. Factors A and B were found by two different methods of assessing factor similarity (inspection and coefficients of congruence) to be highly similar across both stimulus model and subject age levels; Factor C (a weakly defined factor) was found to be similar only for the older stimulus models across subject age levels. An examination of the scales heavily loaded on each factor for all relevant factor matrices resulted in the interpretation of Factors A, B, and C as reflecting Interpersonal Ability, Instrumental Ability, and Propriety, respectively. This study resulted in findings bearing upon three interrelated areas. First, the data suggest that: 1) older and younger adults constitute qualitatively approximately the same class of stimuli, and 2) older and younger adults constitute qualitatively approximately the same population of subjects. That is, perceptions were based on the same underlying dimensions of meaning regardless of age. Second, the data suggest that two major dimensions of social evaluation are Interpersonal Ability and Instrumental Ability. Comparisons of these two attitudinal dimensions with dimensions from other semantic differential studies revealed striking similarity. Third, the finding of stable scales and replicable dimensions suggests the appropriateness of constructing a cross-age semantic differential for social evaluation. Consequently, a 15-scale Cross-Age Semantic Differential for the measurement of social evaluation was proposed.

No description available.

No description available.