Jusqu'à l'avènement de méthodes de neuroimagerie non invasives les connaissances du cerveau sont acquis par l'étude de ses lésions, des analyses post-mortem et expérimentations invasives. De nos jours, les techniques modernes d'imagerie telles que l'IRMf sont capables de révéler plusieurs aspects du cerveau humain à une résolution spatio-temporelle progressivement élevé. Cependant, afin de pouvoir répondre à des questions neuroscientifiques de plus en plus complexes, les améliorations techniques dans l'acquisition doivent être jumelés à de nouvelles méthodes d'analyse des données. Dans cette thèse, je propose différentes applications de l'apprentissage statistique au traitement des données d'IRMf. Souvent, les données acquises par le scanner IRMf suivent une étape de sélection de variables dans lequel les cartes d'activation sont extraites du signal IRMf. La première contribution de cette thèse est l'introduction d'un modèle nommé Rank-1 GLM (R1-GLM) pour l'estimation jointe des cartes d'activation et de la fonction de réponse hémodynamique (HRF). Nous quantifions l'amélioration de cette approche par rapport aux procédures existantes sur différents jeux de données IRMf. La deuxième partie de cette thèse est consacrée au problème de décodage en IRMf, ce est à dire, la tâche de prédire quelques informations sur les stimuli à partir des cartes d'activation du cerveau. D'un point de vue statistique, ce problème est difficile due à la haute dimensionnalité des données, souvent des milliers de variables, tandis que le nombre d'images disponibles pour la formation est faible, typiquement quelques centaines. Nous examinons le cas où la variable cible est composé à partir de valeurs discrets et ordonnées. La deuxième contribution de cette thèse est de proposer les deux mesures suivantes pour évaluer la performance d'un modèle de décodage: l'erreur absolue et de désaccord par paires. Nous présentons plusieurs modèles qui optimisent une approximation convexe de ces fonctions de perte et examinent leur performance sur des ensembles de données IRMf. Motivé par le succès de certains modèles de régression ordinales pour la tâche du décodage basé IRMf, nous nous tournons vers l'étude de certaines propriétés théoriques de ces méthodes. La propriété que nous étudions est connu comme la cohérence de Fisher. La troisième, et la plus théorique, la contribution de cette thèse est d'examiner les propriétés de cohérence d'une riche famille de fonctions de perte qui sont utilisés dans les modèles de régression ordinales. / Until the advent of non-invasive neuroimaging modalities the knowledge of the human brain came from the study of its lesions, post-mortem analyses and invasive experimentations. Nowadays, modern imaging techniques such as fMRI are revealing several aspects of the human brain with progressively high spatio-temporal resolution. However, in order to answer increasingly complex neuroscientific questions the technical improvements in acquisition must be matched with novel data analysis methods. In this thesis we examine different applications of machine learning to the processing of fMRI data. We propose novel extensions and investigate the theoretical properties of different models. % The goal of an fMRI experiments is to answer a neuroscientific question. However, it is usually not possible to perform hypothesis testing directly on the data output by the fMRI scanner. Instead, fMRI data enters a processing pipeline in which it suffers several transformations before conclusions are drawn. Often the data acquired through the fMRI scanner follows a feature extraction step in which time-independent activation coefficients are extracted from the fMRI signal. The first contribution of this thesis is the introduction a model named Rank-1 GLM (R1-GLM) for the joint estimation of time-independent activation coefficients and the hemodynamic response function (HRF). We quantify the improvement of this approach with respect to existing procedures on different fMRI datasets. The second part of this thesis is devoted to the problem of fMRI-based decoding, i.e., the task of predicting some information about the stimuli from brain activation maps. From a statistical standpoint, this problem is challenging due to the high dimensionality of the data, often thousands of variables, while the number of images available for training is small, typically a few hundreds. We examine the case in which the target variable consist of discretely ordered values. The second contribution of this thesis is to propose the following two metrics to assess the performance of a decoding model: the absolute error and pairwise disagreement. We describe several models that optimize a convex surrogate of these loss functions and examine their performance on different fMRI datasets. Motivated by the success of some ordinal regression models for the task of fMRI-based decoding, we turn to study some theoretical properties of these methods. The property that we investigate is known as consistency or Fisher consistency and relates the minimization of a loss to the minimization of its surrogate. The third, and most theoretical, contribution of this thesis is to examine the consistency properties of a rich family of surrogate loss functions that are used in the context of ordinal regression. We give sufficient conditions for the consistency of the surrogate loss functions considered. This allows us to give theoretical reasons for some empirically observed differences in performance between surrogates.

In digital communication, information is sent as bits, which is corrupted by the noise present in wired/wireless medium known as the channel. The Low Density Parity Check (LDPC) codes are a family of error correction codes used in communication systems to detect and correct erroneous data at the receiver. Data is encoded with error correction coding at the transmitter and decoded at the receiver. The Noisy Gradient Descent BitFlip (NGDBF) decoding algorithm is a new algorithm with excellent decoding performance with relatively low implementation requirements. This dissertation aims to characterize the performance of the NGDBF algorithm. A simple improvement over NGDBF called the Re-decoded NGDBF (R-NGDBF) is proposed to enhance the performance of NGDBF decoding algorithm. A general method to estimate the decoding parameters of NGDBF is presented. The estimated parameters are then verified in a hardware implementation of the decoder to validate the accuracy of the estimation technique.

This thesis examines the identity of the U.S. military personnel (the Marines) through their perceptions of a war television series, Generation Kill. The thesis sampled 38 respondents through an online questionnaire. The thesis uses the encoding and decoding theory to reveal that the Marine respondents demonstrated multiple interpretations of themes and concepts in the official discourses. The negotiated decoding from the respondents shows an interchangeable nature of the concepts that the official discourses and the informal conducts are integral to each other. The thesis makes use of the details in GK to draw out the grunts’ experience of being a Marine. In the end, it shows the Marine identity experienced by individual members is diverse and multilayered, and the identity could be shaped and negotiated. The research not only contributes to the missing research on professional subcultures in audience reception studies, but also engages an interdisciplinary exploration on issues about the military in democratic societies.

No description available.

No description available.

The employment of virtual reality (VR) in the cultural heritage context has been recognized as a new trend shaping the way audiences interact with museums. This research explores the effect of VR experience on art interpretation through an audience reception analysis of two VR exhibitions: Mona Lisa: Beyond the Glass and Dreams of Dalí. An ethnographic method was used, i.e., semi-structured interviews with 10 participants served to understand how audiences react to VR. Hall’s encoding/decoding model and Norman’s concept of affordances were employed to determine whether, and in what way, decoding of the VR affordances affects the interpretation of the presented art. First, the results showed that all five previously identified affordances (immersion, presence, empathy, embodiment and usability) are existent in the VR exhibitions. Space perception as an additional affordance has been identified, including seclusion as an anti-affordance specific to VR exhibitions. Second, the results of the conducted interviews’ analysis show proportionality between the VR affordances and interpretation, i.e., the preferred reading of the artwork. In most cases, the stronger the affordances are, the stronger the comprehension, and the more positive the evaluation. Empathy affordance has demonstrated to be the strongest indicator of a preferred interpretation since it is proportionate to the evaluation of all 10 interviewees. These results have implications both in and outside the cultural heritage context – VR employs unconventional and persuasive modes of communicating; it encourages creative, more inclusive, and tacit learning.

This thesis comprises of two parts. In the first, we improve the performance of existing hybrid FSO/RF communication systems. Conventional hybrid RF and optical wireless communication systems make use of independent and parallel Free Space Optical (FSO) and RF channels to achieve higher reliability than individual channels. This thesis is based on the idea that true hybridization can be accomplished only when both channels collaboratively compensate the shortcomings of each other and thereby, improve the performance of the system as a whole. We believe that optimization on the combined channel capacities instead of the individual channel capacities of the FSO and RF channels can increase the system availability by a large amount. Using analysis and simulation, we show that, by using Hybrid Channel Codes, we can obtain more than two orders of magnitude improvement in bit error rates and many-fold increase in system availability over the currently existing best systems. Simulations also show that the average throughput obtained using the new system is over 35% better when compared to the present systems. The goodput is much higher because of the elimination of data repetition. Also by avoiding data duplication, we preserve to a great extent the crucial security benefits of FSO communications. The second half of the thesis deals with magnetic recording systems. Due to the insatiable and ever-increasing needs of data storage, novel techniques have to be developed to improve the capacity of magnetic recording channels. These capacity requirements translate to improving storage densities and using higher recording rates. For these channels, improvements even in the order of a tenths of a dB have a big impact on the storage densities of the recording device. Recently, LDPC codes have been constructed to achieve the independent and uniformly distributed (i.u.d.) capacity of partial response (PR) channels. The “guess algorithm” has been proposed for memoryless channels, to improve the performance of iterative belief propagation decoding to that of Maximum Likelihood (ML) decoding. In the second part of this thesis, the “guess algorithm” is extended to channels with memory. It is shown using asymptotic density evolution analysis that the gains obtained using this algorithm on these channels are more than those obtained over memoryless channels. The “guess algorithm” is further extended to magnetic recording channels which are characterized by ISI and additive white gaussian noise (AWGN). Simulations show that gains of upto one dB are possible on magnetic recording channels.

Reading is not an easily learned skill for most students. I chose to look at the methodology being used by teachers in East Tennessee to instruct emerging readers. Through my review of literature, I researched reading approaches implemented in American classrooms in the last 150 years. I compared and contrasted data to determine current researchers' findings concerning the most effective techniques for teaching reading and how teachers have implemented this knowledge base into their teaching strategies. Teachers cannot teach what they do not know. Therefore, I also researched literature addressing the growing concern among educators with teachers' preparation and professional development opportunities as well as the amount of specific preparation teachers received regarding the reading methods they are using. After compiling the data from my interviews with 30 East Tennessee first-grade teachers, I found that most said they did not feel adequately prepared to teach emergent readers. These teachers reported they had not had instruction that provided foundations in a wide range of research-based approaches to reading. The professional development offerings for teachers already in the classroom were often sporadic and did not compensate for their lack of preparation in college. There is a need for colleges and universities to re-evaluate the current teacher preparation programs. School systems should strive to provide quality inservice opportunities for instructors of emergent readers as well as hiring reading coaches or specialists to assist the reading instructors.

Studien undersöker hur vårdnadshavare till tonåringar mellan 15-18 år ser på kampanjmaterial från IQ och Systembolaget, vidare hur detta formar konversationerna från vårdnadshavarnas perspektiv om ämnet alkohol. Studien utförs i tre intervjuer i form av fokusgrupper med vårdnadshavare till tonåringar. I intervjuerna fick deltagarna ta del av fem olika kampanjer från både Systembolaget och IQ som är riktade till just vårdnadshavare kring ämnet alkohol och minderåriga. De teoretiska ramverk i studien är encoding/decoding av Stuart Hall (1973), Risksamhället av Ulrich Beck (2009) samt Tvåstegshypotesen av Lazarsfeld och Katz (1955). Målet är att se om dessa kampanjer når ut till den tänkta målgruppen, om kampanjerna används som stöd i konversationerna mellan vårdnadshavare och tonåringar samt vilka faktorer som spelar en roll i konversationerna kring alkohol. Analysmetoden som används är tematisk efter Braun och Clarkes (2006). Där ska vi generalisera mönster och hitta teman i svaren. Resultatet vi fick fram genom studien visar att kampanjmaterialet som finns är tillgängligt om behovet finns. De använder materialet till en viss del som stöd i konversationerna och de anser att det är en efterfrågestyrd kommunikation.

Service delivery and the access to specialized instructions to consumers, encounters many barriers within the profession of speech-language pathology. This state of affairs is largely due to the disparate distribution of speech language services (ASHA, 2005). This restricted access, or an inability to access services, is a result of a number of factors, which include lack of clinicians, insufficient number of facilities in geographic area, and transportation issues (ASHA, 2004e). As a result, students who require specialized reading instruction are not afforded the opportunity to access the necessary treatment. It is essential that the literacy needs of all children be addressed, including those who require specialized instruction (Foorman & Torgesen, 2001; Allington, 1994). Technology, specifically telepractice, is a potential solution to address this dilemma. The purpose of this study is to investigate the reliability and validity of systematic multisensory reading treatment for students who have been identified with a delay in foundational reading skills, addressing foundational reading skills via an internet-based video conferencing system. The results will establish the groundwork for the efficacy, reliability, and validity of internet-based video conferencing as a means of service delivery for foundational reading skills. The foundational reading skills targeted in this study are letter naming knowledge (LNK), letter sound knowledge (LSN) and decoding.