Clustering data by identifying a subset of representative examples is important for detecting patterns in data and in processing sensory signals. Such "exemplars" can be found by randomly choosing an initial subset of data points as exemplars and then iteratively refining it, but this works well only if that initial choice is close to a good solution. This thesis describes a method called "affinity propagation" that simultaneously considers all data points as potential exemplars, exchanging real-valued messages between data points until a high-quality set of exemplars and corresponding clusters gradually emerges. Affinity propagation takes as input a set of pairwise similarities between data points and finds clusters on the basis of maximizing the total similarity between data points and their exemplars. Similarity can be simply defined as negative squared Euclidean distance for compatibility with other algorithms, or it can incorporate richer domain-specific models (e.g., translation-invariant distances for comparing images). Affinity propagation’s computational and memory requirements scale linearly with the number of similarities input; for non-sparse problems where all possible similarities are computed, these requirements scale quadratically with the number of data points. Affinity propagation is demonstrated on several applications from areas such as computer vision and bioinformatics, and it typically finds better clustering solutions than other methods in less time.

This dissertation examines the production, perception and processing of the Catalan-specific mid-vowel categories (/e/-/[open-mid front unrounded vowel]/ and /o/-/[open-mid back rounded vowel]/) by early Spanish-Catalan bilinguals in Majorca (Spain). The analyses focus on the lexical as well as the segmental levels to analyze cognate effects in the production and lexical representations of these early bilinguals, and they explore how their production and perception abilities are related. This study provides evidence that early and highly proficient Spanish-Catalan bilinguals in Majorca maintain two independent phonetic categories in the Catalan mid-vowel space. The first significant finding is that production patterns in Majorca differ from those previously reported in Barcelona, as the Catalan mid-vowel contrasts are not merging into a single Spanish-like mid-vowel for either Catalan-dominants or Spanish-dominants. Additionally, these bilinguals are not 'deaf' to the Catalan-specific mid-vowel contrasts: both language dominance groups perceive the contrast between the Catalan mid-vowel categories despite the overlap with one phonetic category in Spanish. Even though Spanish-dominant bilinguals as a whole are indistinguishable from Catalan-dominant bilinguals in the perception and production tasks, they are found to have a higher error rate in the lexical decision task. The comparison of the acoustic properties of the target vowels in Catalan cognate and non-cognate experimental items reveals that the production of the mid-vowels is affected by cognate status, and that these cognate effects are also found in the word recognition of aurally presented stimuli. Finally, bilinguals who produced the mid-vowels with a smaller Euclidean distance are more likely than bilinguals who maintain a more robust contrast in their productions to have a higher error rate in the AXB discrimination and lexical decision tasks. The present study contributes to the discussion regarding the organization of early bilinguals' dominant and non-dominant phonetic systems, and implications are considered for cross-linguistic models of bilingual speech production and perception. It is proposed that the exemplar model of lexical representation (Bybee, 2001; Pierrehumbert, 2001) can be extended to include bilingual lexical connections that can account for the interactions between the phonetic and lexical levels of early bilingual individuals. / text

No description available.

This thesis investigates the boundary of human rationality and how psychological processes interact with underlying regularities in the environment and affect beliefs and achievement. Two common modes in everyday experiential learning, supervised and unsupervised learning were hypothesized to tap different ecological and epistemological approaches to human adaptation; the Brunswikian and the Gibsonian approach. In addition, they were expected to be differentially effective for achievement depending on underlying regularities in the task environment. The first approach assumes that people use top-down processes and learn from hypothesis testing and external feedback, while the latter assumes that people are receptive to environmental stimuli and learn from bottom-up processes, without mediating inferences and support from external feedback, only exploratory observations and actions. Study I investigates selective supervised learning and showed that biased beliefs arise when people store inferences about category members when information is partially absent. This constructivist coding of pseudo-exemplars in memory yields a conservative bias in the relative frequency of targeted category members when the information is constrained by the decision maker’s own selective sampling behavior, suggesting that niche picking and risk aversion contribute to conservatism or inertia in human belief systems. However, a liberal bias in the relative frequency of targeted category members is more likely when information is constrained by the external environment. This result suggests that highly exaggerated beliefs and risky behaviors may be more likely in environments where information is systematically manipulated, for example when positive examples are highlighted to convey a favorable image while negative examples are systematically withheld from the public eye. Study II provides support that the learning modes engage different processes. Supervised learning is more accurate in less complex linear task environments, while unsupervised learning is more accurate in complex nonlinear task environments. Study III provides further support for abstraction based on hypothesis testing in supervised learning, and abstraction based on receptive bottom-up processes in unsupervised learning that aimed to form ideal prototypes as highly valid reference points stored in memory. The studies support previous proposals that integrating the Brunswikian and the Gibsonian approach can broaden the scope of psychological research and scientific inquiry.

This investigation evaluated the use of multiple exemplar instruction (MEI) to teach perspective-taking skills to adolescents with high-functioning autism. The first experiment used a multiple probe design to examine the use of MEI to teach participants to respond appropriately to the deictic frames of I-You, Here-There, and Now-Then. Participants were instructed and tested using developmentally appropriate perspective-taking protocols, and generalization of perspective-taking skills to a more natural social interaction format was also assessed. The second experiment used a multiple probe design to examine the use of MEI to facilitate defusion from negative thoughts and feelings related to social interaction. Participants were trained to discriminate between different aspects of their own perspective (i.e., self-as-content and self-as-context). Defusion was measured using ratings of the believability and comfort associated with negative thoughts, and the percentage of time engaged in appropriate social interaction was recorded for each participant. Results from Experiment 1 showed the emergence of deictic relational responding in the presence of novel relations for all participants following MEI, and varying degrees of generalization of perspective-taking skills to a natural presentation of social interaction were observed based on the complexity of the deictic relation. Results from Experiment 2 revealed a decrease in believability and an increase in comfort associated with problematic thoughts for all participants following MEI in the absence of directly observed changes in social interaction in the natural environment. These results support the use of MEI as an instructional strategy for teaching perspective-taking skills to individuals with autism. Findings are discussed according to a Relational Frame Theory analysis of perspective-taking.

SPOP paired with multiple exemplar instruction has been shown to be effective with typically developing preschoolers in establishing the joint stimulus control required for the development of naming (Rosales et al., 2012). The purpose of the current investigation is to evaluate the effectiveness and efficiency of the stimulus pairing observation procedure in establishing speaker and listener responses with children with developmental disabilities. Participants were presented with pairings of an auditory and visual stimulus. Testing was conducted after the instructional phase to assess if the participant could tact the item when presented with the picture as well as select the object from an array when given the name of the object. If the participants did not meet mastery criteria during posttests, a multiple exemplar procedure (MEI) was implemented to examine its efficacy in establishing the joint stimulus control of the speaker and listener responses. Following multiple exemplar instruction, testing was conducted. The results of the current investigation show that the SPOP in combination with MEI was effective in establishing both the listener and tact relations for all three participants.

CoordenaÃÃo de AperfeiÃoamento de Pessoal de NÃvel Superior / Este estudo teve por objetivo geral refletir sobre a palatalizaÃÃo das oclusivas alveolares /t, d/ → [tʃ, dʒ], caracterÃstico de muitos falares do PortuguÃs Brasileiro (PB), na construÃÃo da fonologia do InglÃs LÃngua Estrangeira (ILE). Baseados nos preceitos da Fonologia de Uso (BYBEE, 2001), do Modelo de Exemplares (PIERREHUMBERT, 2001) e na visÃo de lÃngua enquanto Sistema Adaptativo Complexo (LARSEN-FREEMAN; CAMERON, 2008), tivemos por hipÃtese bÃsica que aprendizes brasileiros apresentam percursos diferenciados de construÃÃo da fonologia do ILE, a depender de seu falar regional. Este estudo foi uma pesquisa experimental, de cunho concomitantemente transversal-longitudinal. Selecionamos dois campos de pesquisa, Fortaleza-CE e MossorÃ-RN, distintos quanto ao falar regional do PB, sendo o primeiro palatalizador das oclusivas alveolares e o segundo nÃo-palatalizador. Os informantes foram controlados ainda quanto ao sexo e ao nÃvel de proficiÃncia no ILE. Selecionamos itens lexicais propÃcios à emergÃncia da palatalizaÃÃo no PB e no ILE, buscando o controle da frequÃncia de ocorrÃncia, do contexto fonotÃtico e da tonicidade silÃbica. Fizemos uso de 5 Experimentos de coletas de dados. No PB, P1 envolveu uma conversa sobre algumas figuras, enquanto P2 utilizou a leitura de diversas frases-veÃculo. No ILE, I1 envolveu a repetiÃÃo de Ãudio distorcido associado a algumas figuras, I2 utilizou a leitura de diversas frases-veÃculo e I3 usou um jogo da memÃria em sua aplicaÃÃo. Os resultados do estudo transversal envolvendo a emergÃncia da palatalizaÃÃo das oclusivas alveolares do PB enfatizaram o carÃter nÃo-categÃrico de fenÃmeno em ambas as regiÃes: a palatalizaÃÃo foi observada no falar do RN, e oclusivas alveolares nÃo-palatalizadas foram observadas no falar do CE. A anÃlise dos dados transversais do ILE apontou: a) a variÃvel origem como fator importante para determinar a maior ou menor emergÃncia da palatalizaÃÃo, aprendizes do CE tenderam a uma maior palatalizaÃÃo; b) o tipo de vozeamento da oclusiva alveolar, com as desvozeadas mais propensas à palatalizaÃÃo; c) o indivÃduo, sujeitos da mesma regiÃo e nÃvel de proficiÃncia realizaram o fenÃmeno de forma distinta, d) a palavra, itens lexicais com a mesma sequÃncia fonotÃtica apresentaram comportamentos divergentes; e e) o tipo fonotÃtico, alguns tipos apresentaram percentuais semelhantes na realizaÃÃo da palatalizaÃÃo em ambas as regiÃes. Algumas variÃveis mostraram influÃncia relativa na realizaÃÃo do fenÃmeno, como o sexo, o nÃvel de proficiÃncia e a tonicidade silÃbica. Por fim, a frequÃncia de ocorrÃncia dos itens lexicais mostrou-se irrelevante na anÃlise de dados transversais do ILE. Em se tratando dos resultados do estudo longitudinal, observamos que os informantes do RN tenderam a uma pequena realizaÃÃo do Ãndice de PalatalizaÃÃo (IP) com o passar das coletas de dados longitudinais. Os informantes do CE presentaram comportamentos distintos. FM1 apresentou grande variaÃÃo em sua realizaÃÃo do IP, com palatalizaÃÃo mais alta que os informantes do RN. FM3 apresentou uma realizaÃÃo relativamente estÃvel de seu IP individual, com forte palatalizaÃÃo, bem mais recorrente que seus colegas. Os dados do estudo longitudinal reforÃaram a relevÃncia de variÃveis como a origem, o indivÃduo e a palavra na emergÃncia da palatalizaÃÃo no ILE de aprendizes brasileiros. Adicionalmente, durante o perÃodo de coleta de dados longitudinais nÃo encontramos indÃcios significativos de reduÃÃo do IP no ILE com o desenrolar das coletas de dados longitudinais. Tendo em vista tais evidÃncias, tomamos por confirmada a hipÃtese bÃsica que aprendizes brasileiros apresentam percursos diferenciados de construÃÃo da fonologia do ILE, a depender de seu falar regional. / This study had as its main objective to reflect upon alveolar stop palatalization /t, d/ → [tʃ, dʒ] phenomenon, characteristic of many Brazilian Portuguese (BP) dialects, in the construction of the English as a Foreign Language (EFL) phonology. Grounded on the ideas proposed by Usage-Based Phonology (BYBEE, 2001), Exemplar Models (PIERREHUMBERT, 2001), as well as on the view of language as a Complex Adaptive System (LARSEN-FREEMAN; CAMERON, 2008), the basic hypothesis stated Brazilian EFL learners follow different construction pathways of the EFL phonology, depending on their regional dialect. This was an experimental research, following both a cross-sectional and longitudinal design. It had two research fields, Fortaleza-CE and MossorÃ-RN, which differed on their BP dialect, the former a palatalizing dialect and the latter a non-palatalizing one. Subjects were also controlled by sex and EFL proficiency level. Lexical items which were susceptible to palatalization in both BP and EFL were selected, aiming to control token frequency, phonotactics and syllable stress. 5 experiments were used to collect data. For BP, P1 involved a picture-induced conversation, while P2 used a carrier-sentence reading procedure. For EFL, I1 involved the repetition of distorted audio tokens associated to pictures, I2 used a carrier-sentence reading procedure and I3 focused on a memory game. BP cross-sectional results relating to the emergence of palatalization in both regions emphasized non-categorical realization of the phenomenon: palatalized tokens were observed in RN and non-palatalized ones were found in CE. EFL cross-sectional data analysis indicated: a) the variable origin as an important factor of bigger or smaller palatalization emergence, as CE learners consistently tended to higher palatalization levels; b) the alveolar stop voicing pattern, as voiceless sounds tended to palatalize more frequently; c) the individual, as subjects from the same area and proficiency level realized the phenomenon with different patterns; d) the word, as lexical items with the same phonotactic structure allowed higher or smaller palatalization emergence; and e)phonotactic structure, as a few words allowed similar palatalization emergence in both study regions. Some variables were not so relevant for the emergence of the phenomenon, like sex, proficiency level and syllable stress. Finally, token frequency was not relevant at all in the EFL cross-sectional data. As regards longitudinal results, it was observed RN subjects tended to a low Palatalization Index (PI) as longitudinal data collection took place. CE subjects had distinct behavior. FM1 had great variation on his PI, with higher PI values than RN informants. FM3 had a relatively stable PI realization, with the highest palatalization level. Longitudinal data reinforced the value of variables such as the origin, the individual and the word on the emergence of EFL palatalization of Brazilian learners. Additionally, significant PI reduction during longitudinal data collection was not observed. Having these evidences in mind, it was concluded Brazilian learners follow through different EFL phonology construction pathways, depending on their regional dialect.

Image inpainting is the process of recovering missing or deteriorated data within the digital images and videos in a plausible way. It has become an important topic in the area of image processing, which leads to the understanding of the textural and structural information within the images. Image inpainting has many different applications, such as image/video restoration, text/object removal, texture synthesis, and transmission error concealment. In recent years, many algorithms have been developed to solve the image inpainting problem, which can be roughly grouped into four categories, partial differential equation-based inpainting, exemplar-based inpainting, transform domain inpainting, and hybrid image inpainting. However, the existing algorithms do not work well when the missing region to be inpainted is large, and when there are textural and structural information needed to be recovered. To address this inpainting problem, we propose multiple algorithms, 1) perceptually aware image inpainting based on the perceptual-fidelity aware mean squared error metric, 2) image inpainting using nonlocal texture matching and nonlinear filtering, and 3) multiresolution exemplar-based image inpainting. The experimental results show that our proposed algorithms outperform other existing algorithms with respect to both qualitative analysis and observer studies when inpainting the missing regions of images.

In the recent decades, alternative notions regarding the role of symbols in intelligence in natural and artificial systems have attracted significant inter- est. The main difference of the so-called situated and embodied approaches to cognitive science from the traditional cognitivist position is that symbolic repre- sentations are viewed as resources, similar to maps used for navigation or plans for activity, instead of as transparent stand-ins in internal world models. Thus, all symbolic resources have to be interpreted and re-contextualized for use in concrete situations. In this view, one of the primary sources of such symbolic resources is language. Cognitivism views language as a vessel carrying informa- tion originally located in the processing mechanisms of the individual agents. Situated approaches, on the other hand, view language both as a communicative mechanism and as a means for the individual agents to enhance and extend their cognitive machinery, by e.g. better utilizing their attentional resources, or mod- ifying their perceptual-motor means. Taking inspiration from these ideas, and building on multi-agent models developed in other fields, the field of language evolution developed models of the emergence of shared resources for communi- cation in a community of agents. In these models, agents with various means of categorization and learning engage in communicative interactions with each other, using shared signs to refer either to pre-given meanings or entities in a situation. In order to avoid falling into the same mentalist pitfalls as cognitivism in the design of these models, such as the stipulation of an inner sphere of mean- ings for which communicative signs are mere labels, the role of communication should be viewed as one of the social coordination of behavior using physically grounded symbols. To this end, an experimental setup for language games, and a robotic model for agents which engage in such games are presented. The setup allows the agents to utilize shared symbols in the completion of a simple task, with one agent instructing another on which action to undertake. The symbols used by agents in the language games are grounded in the embodied choices presented to them by their environment, and the agents can further use the symbols created in these games for enhancing their own behavioral means. The learning mechanism of the agents is similarity-based, and uses low-level sensory data to avoid the building in of features. Experiments have shown that the establishment of a common vocabulary of labels depends on how well the instructors are trained on the task and the availability of feedback mechanisms for the exchanged labels.

ABSTRACT  The aim of this thesis is to explore notions of spiritual leadership in a church environment and how conceptions of spiritual leadership are discussed and presented by the three different authors, all clergy and scientists, Jackson W. Carrol, Tone Stangeland Kaufmann and Karin Johannesson.  The conceptions of spiritual leadership is analysed through the theoretical perspective provided by Linda Trinkaus Zagzebski in Exemplarist Moral Theory. Spiritual leadership is discussed as a conception based on both an inner experience and behaviour in relation to its social and societal environment. According to the findings in this thesis, the virtue ethics of Exemparist Moral Theory provides opportunities to propose a suitable conception of spiritual leadership based on the use of exemplars from the biblical narrative, making active use of Scripture and its suggestions of egalitarian behaviour. It also presents some challenges discussing how inner and subjective experiences of spirituality manifests itself in actions, how it is to be regarded from an egalitarian perspective plus a recurring question of whether the presented conceptions are exceptional enough to stimulate emotions and be consequently admired.  The findings in this thesis, through the perspective of Exemplarist Moral Theory, therefore indicates a certain direction for spiritual leaders in church which is to act virtuous in accordance with the biblical narrative, with Jesus Christ as its foremost exemplar. Keywords: Spiritual leadership, exemplar, virtue, Jackson W. Carrol, Tone Stangeland Kaufmann, Karin Johannesson.