Les travaux de recherche développés dans cette thèse, définissent une approche d'évolution d'ontologie Onto-Evoal (Ontology Evolution-Evaluation) qui s'appuie sur une modélisation de patrons de gestion de changement CMP (Change Management Patterns). Ces patrons spécifient des classes de changements, des classes d'incohérences et des classes d'alternatives de résolution. Sur la base de ces patrons et des relations sémantiques entre eux, un processus automatisé permettant de conduire l'application des changements tout en maintenant la cohérence de l'ontologie évoluée a été développé. L'approche intègre également une activité d'évaluation basée sur un modèle de qualité d'ontologie qui a été défini. Ce modèle est employé pour guider la gestion des incohérences en évaluant l'impact des résolutions proposées sur le contenu et l'usage de l'ontologie à travers un ensemble de métriques quantitatives et ce, afin de choisir une résolution qui préserve la qualité de l'ontologie évoluée. La gestion des changements étant fortement liée au modèle dans lequel est représentée l'ontologie, nous nous sommes focalisés sur le langage OWL en tenant compte de l'impact des changements sur la cohérence logique de l'ontologie telle que spécifiée dans la couche OWL DL. Les principales contributions de l'approche résident dans la modélisation des patrons de gestion de changement guidant le processus d'évolution, l'intégration de l'évaluation de la qualité pour optimiser la résolution des changements et la modélisation formelle et explicite du journal d'évolution.

CoordenaÃÃo de AperfeiÃoamento de Pessoal de NÃvel Superior / O presente estudo à uma investigaÃÃo que tem como objetivo discutir as estratÃgias evidenciais identificadas em seus aspectos sintÃticos, semÃnticos e pragmÃtico-discursivos no material didÃtico impresso para cursos superiores da modalidade a distÃncia, com o objetivo de avaliar as indicaÃÃes das fontes dos conteÃdos asseverados. O corpus foi constituÃdo de oito livros, escolhidos conforme Tabela de Ãreas do Conhecimento (CNPQ), distribuÃdos da seguinte forma: quatro da grande Ãrea CiÃncias Humanas (EducaÃÃo, Antropologia, Psicologia e Filosofia) e quatro da grande Ãrea LinguÃstica, Letras e Artes (LÃngua Portuguesa, LinguÃstica II, HistÃria das Artes Visuais I, LaboratÃrio de Teatro I). Considerando que o interesse deste estudo à a competÃncia comunicativa dos falantes envolvidos no processo ensino-aprendizagem, foi utilizado como aporte teÃrico o Funcionalismo LinguÃstico, mais especificamente a GramÃtica Discursivo-Funcional (GDF), desenvolvida por Hengeveld e Mackenzie (2008). Como resultado da anÃlise, verificamos, entre outros, os seguintes aspectos: que as estratÃgias evidenciais no material didÃtico impresso para cursos superiores da modalidade a distÃncia apresentam aspectos sintÃticos, semÃnticos e pragmÃtico-discursivos que indicam como tipo de fonte e como tipo de evidencialidade predominantes, respectivamente, uma terceira pessoa definida e evidencialidade reportativa, demonstrando um distanciamento dos autores em relaÃÃo aos conteÃdos, tendo em vista o uso do chamado âargumento de autoridadeâ no embasamento teÃrico dos textos; que a evidencialidade dos tipos inferida e genericidade ocorrem, mas nÃo sÃo frequentes e o tipo percepÃÃo de evento nÃo foi encontrado, uma vez que os conteÃdos apresentados nesse tipo de produÃÃo textual sÃo mais conceituais, isto Ã, assumem uma natureza mais abstrata; que, quanto ao modo de acesso, hà uma predominÃncia do relato de 2 mÃo, o que nÃo nos surpreende, uma vez que foram predominantes a evidencialidade reportativa e a fonte do tipo terceira pessoa definida; que as marcas linguÃsticas de adesÃo do enunciador à fonte indicada em relaÃÃo a um determinado conteÃdo reportado nesse material ocorrem de forma pouco significativa, porÃm, quando utilizadas, constituem-se de adjetivos qualificativos da fonte, verbos e outros, como advÃrbios e operadores argumentativos; que o produtor do material didÃtico analisado prefere nÃo demonstrar algum tipo de refutaÃÃo/afastamento à fonte da informaÃÃo; que nÃo à frequente a coocorrÃncia de estratÃgias linguÃsticas da modalidade epistÃmica na estratÃgia evidencial usada na indicaÃÃo da fonte dos conteÃdos asseverados, mas, quando utilizadas produzem um baixo comprometimento; que o verbo à o meio linguÃstico mais usado para manifestar a evidencialidade; que houve uma predominÃncia pela anteposiÃÃo da marca evidencial à fonte e ao conteÃdo; que a fonte do tipo enunciador està ligada, mais frequentemente, à evidencialidade inferida, enquanto a fonte do tipo terceira pessoa definida està ligada, mais comumente, à evidencialidade reportativa e as fontes do tipo domÃnio comum e terceira pessoa indefinida se ligam, mais comumente, à evidencialidade genericidade; que os tipos evidenciais reportativa e genericidade promovem efeito de baixo comprometimento, enquanto o tipo inferida està ligada, mais comumente, a um mÃdio comprometimento; que a marca de adesÃo serve para indicar um maior comprometimento do produtor textual em relaÃÃo ao conteÃdo asseverado; que, quando a fonte à o enunciador, o efeito à de um mÃdio comprometimento, tendo em vista que o produtor de material didÃtico impresso para EaD utiliza a primeira pessoa do plural, tal como no chamado âplural de modÃstiaâ; que, quando a fonte à de terceira pessoa indefinida, hà um equilÃbrio entre o mÃdio e baixo comprometimento; que o uso da fonte terceira pessoa definida indica, em geral, um baixo comprometimento do produtor textual; e que os meios estÃo relacionados, mais comumente, ao baixo comprometimento. / The present study is an investigation that aims to discuss the evidential strategies identified in their syntactic, semantic and pragmatic- discursive in printed courseware for higher education courses in the distance mode, with the aim of evaluating the indications of the sources of ensured content. The corpus was composed of eight books, chosen according to the Table of Knowledge Areas (CNPq), distributed as follows: four in the large Area of Humanities ( Education, Anthropology, Psychology and Philosophy) and four in the large Area of Linguistics, Letters and Arts ( Portuguese Language, Linguistics II, History of Visual Arts I, Theatre Lab I). Whereas the interest of this study is the communicative competence of speakers involved in the teaching-learning process, was used as the theoretical functionalism Language, specifically Functional Discourse Grammar (FDG), developed by Hengeveld and Mackenzie (2008). As a result of the analysis, we find, among others, the following: the evidential strategies in printed courseware for higher education students in distance mode have syntactic, semantic and pragmatic-discursive indicate as font type and how to type evidentially predominant, respectively, and a third person defined evidentially reportative, showing a gap of authors in relation to the contents, considering the use of so-called" argument from authority" in theoretical texts; evidentially that the inferred types and genericity occur but are not frequent, and the type of event perception was not found, since the contents in this type of writing are more conceptual, that is, assume a more abstract nature and that, on the access mode, there is a predominance of reports of 2nd hand, which is not surprising, since they were predominantly from evidentially reportative and source type third person set, the linguistic marks of accession of enunciating the source indicated in relation to a particular content reported in this material occur in insignificant, however, when used constitute the source of qualifying adjectives, verbs and others, such as adverbs and argumentative operators, the producer of textbooks analyzed prefers not to show any kind of rebuttal/out the source of the information, which is not frequent co-occurrence of linguistic strategies of epistemic modality evidential strategy used in the source of the ensured content and when used produce a low commitment, the verb is the medium most used language to express evidentiality, there was a predominance ante position by brand evidential the source and content, the source type annunciator is on, most often, the evidentiality inferred as the source type defined third person is linked most commonly to evidentiality reportative and sources of the type common domain and third person undefined bind, more commonly, the evidentiality generality, that the types and genericity evidential reportative promoting effect of low commitment, while the inferred type is linked most commonly to an average commitment; membership mark that serves to indicate a greater commitment producer in relation to textual content asserted, that when the source is the utterer, the effect is of a medium commitment, considering that the producer of printed courseware for DL uses the first person plural, as in so-called " plural of modesty, "that when the source is indefinite third person there is a balance between the medium and low commitment, that the use of the third person defined source indicates, in general, a low textual commitment of the producer, and that means are related, most commonly, low commitment.

nÃo hà / Apesar das vÃrias opÃÃes de ambientes virtuais de aprendizagem no cenÃrio educacional, hà pouco resultado nas aÃÃes que ensejam uma maior motivaÃÃo e envolvimento dos alunos numa utilizaÃÃo mais eficiente destes. Os ambientes virtuais de aprendizado tÃm se apresentado com configuraÃÃes semelhantes no decorrer dos anos. Como forma de envolver os alunos, e buscar resultados mais efetivos nestes ambientes, à introduzida neste trabalho, uma nova arquitetura de ambiente virtual de aprendizagem, baseado nas tecnologias web 2.0 no sentido da personalizaÃÃo em massa. Em um contexto educacional a personalizaÃÃo em massa à um conceito onde o prÃprio usuÃrio tem maior autonomia na decisÃo da forma e conteÃdo do seu ambiente. Jà o termo web 2.0 refere-se as novas aplicaÃÃes web que proporcionam uma rica experiÃncia ao usuÃrio, com recursos de Ãudio, vÃdeo e principalmente diversas possibilidades de interaÃÃo, enxergando o usuÃrio como parte integrante do sistema, muitas vezes atà como o principal criador da informaÃÃo. Neste trabalho foi definido um modelo de arquitetura, com o objetivo de mostrar possÃveis aplicaÃÃes pedagÃgicas com essas novas tecnologias, bem como uma definiÃÃo de requisitos baseado na engenharia de software. Analisaram-se ferramentas e recursos que devem compor este novo ambiente para web 2.0 e com mÃltiplas possibilidades de personalizaÃÃo, das fontes de informaÃÃo aos aplicativos, como tambÃm os fatores que fazem essas tecnologias propÃcias à prÃtica educacional, como tambÃm questÃes relativas ao comportamento do aluno. TambÃm foram analisados trabalhos que apresentavam propostas relacionadas e exemplos destas tecnologias na web atual. Novas formas de comunicar e aprender desafiam as instituiÃÃes de ensino, um ambiente baseado nestes conceitos nÃo à constituÃdo apenas de novas tecnologias, mas sim de uma quebra de paradigma para um ambiente de aprendizagem com menor nÃvel de controle, onde o aluno possui maior liberdade e domÃnio da informaÃÃo. / Though several options of virtual learning environments in the educational scenario there is a few result in actions wich crave greater motivation and involvement of the students in a use more efficient of these. The virtual learning environments have been presented with similar settings over the years. Used as a way to involve the students and reach more effective results in these environments is introduced in this paper a new architecture of virtual learning environments based on Web 2.0 technologies towards mass customization. Considering an educational context the mass customization is a concept where the user has more autonomy in deciding about the form and the content of his own environment. Already the term Web 2.0 refers to the new web applications that provide a great experience to the user with resources of audio, video and mainly many possibilities of interaction. The user is viewed as part of the system and even the main creator of the information. An architecture model was defined in order to show possible pedagogical applications with these new technologies as well as a definition of requirements based on software engineering. Tools and resources were analysed and they must compose this new environment for web 2.0 with multiple possibilities for customization from the sources of information to applications, as well as the factors that make these technologies conducive to educational practice, questions related to the behavior of the student and some analyses of other studies that showed proposals and examples of these technologies on the todayÂs web. New ways to communicate and learn defy the institutions of education, an environment based on these concepts is not composed only of new technology but even about a breach of paradigm to a learning environment with lower level of control where the student has more freedom and field of information.

Libraries have for hundreds of years been an important entity for every kind of institute, especially in the educational sector. So now it is an age of computers and internet. People are now using electronic resources to fulfill their needs and requirements of their life. Therefore libraries have also converted to computerized systems. People can access and use library resources just sitting at their computers by using the internet. This modern way of running a library has been called or given the name of digital libraries. Digital libraries are getting famous for flexibility of use and because more users can be facilitated at a time. As numbers of users are increasing, some issues relevant to interaction also arise while using digital libraries interface and utilizing its e-resources. In this thesis we evaluate usability factors and issues in digital libraries and the authors have taken as a case study the real time existing system of the digital library in BTH. This thesis report describes digital libraries and how users are being facilitated by them. Usability issues are also discussed relevant to digital libraries. Users have been the main source to evaluate and judge usability issues while interacting and using this digital library. The results obtained showed dis¬satisfaction of users regarding the usability evaluation of BTH:s digital library. The authors used usability evaluation techniques to evaluate functionality and services provided by the BTH digital library system interface. Moreover, based on the results of our case study, suggestions of improvement in BTH:s digital library are presented. Hopefully, these suggestions will help to make BTH digital library system more usable in an efficient and effective manner for users. / 0046-738956073 0046-738956073 0046-734956502

Lorsqu’ils travaillent en équipe, les humains ont rarement des comportements optimaux : ils peuvent faire des erreurs, manquer de motivation ou de compétence. Dans les domaines des environnements virtuels ou des systèmes multi-agents, de nombreux travaux ont cherché à reproduire les comportements d’équipes humaines : un agent représente alors un membre de l’équipe. Cependant, ces travaux ont très souvent pour objectif la performance de l’équipe, et non la fidélité des comportements produits. Pour former un apprenant en environnement virtuel à prêter attention et à s’adapter aux autres, nous avons cherché dans cette thèse à reproduire des comportements humains réalistes et non-optimaux de travail d’équipe. Plus particulièrement, nous nous sommes intéressés aux équipes auto-organisées, c’est-à-dire aux équipes dans lesquelles le pouvoir de décision est réparti entre les membres, et dans lesquelles l’organisation est implicite. Dans de telles équipes, l’organisation se fait non pas au travers des communications mais par l’observation et l’anticipation des comportements des autres. Pour s’organiser, chaque agent doit se demander ce qu’il est préférable de faire en fonction de ce que pourraient faire les autres, et donc se poser des questions telles que Ai-je confiance en la compétence de mon coéquipier pour faire cette tâche ? Les relations de confiance permettent donc à chacun de prendre en compte les autres. Pour générer de tels comportements, nous proposons un système permettant aux agents de raisonner d’une part sur un modèle de l’activité à effectuer et d’autre part sur les relations de confiance qui les lient aux autres agents de l’environnement. Dans ce cadre, notre première contribution porte sur l’augmentation du langage de description de l’activité Activity-DL de manière à permettre la description d’activités collectives. Nous proposons également des mécanismes de propagation de contraintes et d’informations qui faciliteront le raisonnement des agents. Ces contraintes et informations permettront par exemple aux agents de répondre à la question Avons-nous les compétences nécessaires pour faire la tâche qui permettra de réaliser notre but collectif ? Notre seconde contribution porte sur la proposition d’un modèle d’agent opérationnalisant le modèle de confiance de Mayer et al. (1995), sélectionné après une étude de la littérature en sciences humaines et sociales sur la confiance. La confiance d’un agent en un autre est décrite selon trois dimensions : l’intégrité, la bienveillance et les compétences. Chaque agent est donc défini par ces trois dimensions et a des croyances sur l’intégrité, la bienveillance et la compétence des autres agents. De plus chaque agent possède des buts qui lui sont personnels ainsi que des buts collectifs et devra donc choisir quel but privilégier. Finalement nous proposons un moteur décisionnel qui permet à chaque agent de calculer l’importance qu’il accorde à ses buts afin de sélectionner une tâche. Nous avons défini les mécanismes de calcul de l’importance des buts de manière à modéliser l’influence sur l’agent de ses croyances sur les autres, et pour sélectionner une tâche, l’agent raisonne à la fois sur les modèles d’activité et sur ses attentes à propos du comportement des autres, également générées à partir des croyances de l’agent sur les autres. Nous avons implémenté notre système et constaté qu’il répond à nos objectifs de génération de comportements d’équipe réalistes et non optimaux. Nous avons également conduit une évaluation perceptive préliminaire au cours de laquelle les participants ont notamment été capables de percevoir la confiance ou le manque de confiance d’un agent en un autre grâce à son comportement. / When working in teams, humans rarely display optimal behaviors: they sometimes make mistakes, lack motivation or competence. In virtual environments or in multi-agent systems, many studies have tried to reproduce human teamwork: each agent acts as a team member. However, the main objective in those studies is the performance of the team: each agent should display optimal behavior, and the realism of those simulated behaviors is not a concern. To train someone in a virtual environment to pay attention to and to adapt to their teammates, we built a decision-making system for agents to display realistic and non-optimal behaviors. More specifically, we are interested in self-organized teams (i.e. teams where the decision power is decentralized among its members) and in implicit organization (i.e. when team members do not interact through communications but rather through the observation of others’ behaviors). In such a team, each agent has to think about what it should do given what others could do. Agents then have to ask themselves questions such as Do I trust my teammate’s competence to perform this task? Trust relationships therefore allow agents to take others into account. We propose a system that allows agents to reason, on the first hand, on models of the activity they have to do, and on the other hand, on trust relationships they share with others. In that context, we first augmented the Activity-Description Language so that it supports the description of collective activities. We also defined mechanisms for constraint generation that facilitates agent reasoning, by giving them the answer to questions like Do we have the required abilities to perform the task which will achieve our goal? We then proposed an agent model based on the model of interpersonal trust of Mayer et al. (1995) that we selected after a study of trust in social science. This model describes trust relationship with three dimensions: the trustor trusts the trustee’s integrity, benevolence and abilities. An agent is therefore defined through those three dimensions, and has a mental model of each other agent; i.e. has trust beliefs about others’ integrity, benevolence and abilities. Moreover each agent has both personal and collective goals (i.e. goals that are shared with other members of the team), and thus will have to decide which goal to focus on. Finally we proposed a decision-making system that allows an agent to compute the importance it gives to its goals and then to select a task. When computing goal importance, the agent is influenced by its trust beliefs about others, and to select a task, it reasons on the activity models and on its expectations about what others could do. Those expectations are generated from the agents’ trust beliefs. We implemented our system and observed that it produces realistic and non-optimal behaviors. We also conducted a preliminary perceptive evaluation which showed that participants were able to recognize one agent’s trust or lack of trust in another through the behaviors of the first one.

<div>Proteins are susceptible to physical and chemical degradation in solution, which can lead to the loss of therapeutic activity and increase the potential for immunogenic responses when administered. Many degradation reactions are mediated by water, and therefore the proteins are often formulated as solids in which degradation rates are slowed significantly. Lyophilization is the most common method for producing solid protein formulations, which removes the water by sublimation and desorption under vacuum from the frozen protein solutions. Lyophilization requires excipients to protect the protein from the inherent stresses involved in the process. Degradation can still occur during lyophilization and storage, and needs to be characterized in order to develop a successful formulation with desired storage stability. The analytical techniques to characterize solid-state proteins are limited, however, and many do not provide site-specific information and lack the ability to predict stability beforehand.</div><div>Recently, solid-state hydrogen-deuterium exchange mass spectrometry (ssHDX-MS) has been developed to characterize proteins in solid powders with peptide level resolution. The technique was found to be sensitive to formulation and process changes. The ssHDX-MS metrics are highly correlated to the long-term storage stability, suggesting that the method can serve as a formulation screening tool. This dissertation aims to evaluate the factors affecting ssHDX kinetics and to develop a mechanistic understanding of the exchange process in solid samples, which in turn will support the solid-state protein development and enable it to be conducted in a more a cost and time-effective way. First, the contribution of peptide-matrix interactions to deuterium incorporation kinetics in the absence of higher-order structure was assessed using lyophilized poly-D, L-alanine peptides. Deuterium incorporation depended on excipient type and D₂O_(g) activity in the solid samples. A reversible pseudo-first-order kinetic model was proposed and validated using the experimental data. Second, the reversibility of the hydrogen-deuterium exchange reaction in the solid-state was evaluated to support the ssHDX mechanistic model further. The reaction was found to be reversible irrespective of initial conditions and independent of the excipient type. Pre-hydration of the peptide samples prior to deuterium labeling did not affect deuterium incorporation in amorphous samples compared to the controls not subjected to pre-hydration. Third, the contribution of peptide secondary structure to deuterium uptake kinetics was quantified using structured PDLA analogs. The deuterium incorporation in structured peptides was less than that of the PDLA peptides suggesting that both peptide structure and peptide-matrix interactions contribute to ssHDX-MS. Finally, a quantitative data analysis method was presented that allows the interpretation of ssHDX-MS data of a protein relative to controls. Altogether, the findings present a comprehensive mechanistic understanding of the ssHDX-MS of proteins that is relevant to the industry.</div>

Organisationer investerar årligen avsevärda summor pengar i ledarskapsutbildningar enligt tidigare forskning. Ledarskapsutbildningen UL är en av Sveriges mest etablerade. Det är av intresse att studera hur utbildningens målsättning uppfylls hos dess deltagare. Denna studie är en kandidatuppsats i Personal och arbetslivsprogrammet på Högskolan Dalarna. Studiens syfte är att undersöka förståelsen för det egna ledarskapet i relation till UL:s målsättningar mellan respondenter som genomgått utbildningen år 2020 respektive 2022. Studiens utgångspunkt grundar sig i ett medarbetarperspektiv men resultatet är även av intresse ur ett arbetsgivarperspektiv. Tolkningen av denna kvalitativa studies empiri grundar sig i den begränsade tidigare forskning som finns kring en av Sveriges mest etablerade ledarskapsutbildningar. Åtta intervjuer med chefer inom kommunal verksamhet ligger till grund för studiens datainsamling. Respondenternas svar har sedan sammanställts via en tematisk analys. Resultatet visar att förståelsen för det egna ledarskapet i relation till UL:s målsättningar skiljer sig mellan respondenter som genomgått utbildningen år 2020 respektive 2022. En mindre skillnad mellan respektive respondentgrupp kunde utläsas gällande upplevelsen av dialog för att uppnå högre effektivitet. I den ena gruppen återger flera respondenter endast vad som bör göras hypotetiskt medan merparten av den andra gruppen ingående beskriver genomförandet knutet till faktiska handlingar och situationer. Respondentgrupperna kunde ej skiljas åt gällande reflektionens betydelse, dock så återgavs tidsbrist i det dagliga arbetet av samtliga respondenterna som ett hinder för utförandet samt prioriterandet av reflektionen. Den markanta skillnaden mellan grupperna återfanns i användandet av självinsikt för att utveckla det egna ledarskapet. Den ena gruppen beskrev aktivt hur de använde sig av självinsikt samt medvetenhet kring det, något som ej kunnat identifierats i den andra gruppen som endast återgav förmodade svar i form av vad som borde göras eller övergripande svar om hur ledare i allmänhet bör agera. / Organizations annually invest significant amounts of money in leadership training, according to previous research. The leadership training UL is one of Sweden´s most established. It is of interest to study how the objectives of the education are met by its participants. This study is a bachelor´s thesis in the Human Resources and Labour Relations Programme at Dalarna University. The purpose of this study is to examine the understanding of one's own leadership in relation to UL's objectives between respondents who completed the education in 2020 and 2022. The study's starting point is based on an employee perspective, but the results are also of interest from an employer perspective. The interpretation of this qualitative empirical study is based on the limited previous research that exists around one of Sweden's most established leadership educations. Eight interviews with managers in municipal operations form the basis for the study's data collection. The respondents' answers have then been compiled via a thematic analysis. The results show that the understanding of one's own leadership in relation to UL's objectives differs between respondents who completed the training in 2020 and 2022, respectively. In one group, several respondents only reproduce what should be done hypothetically, while most of the other group are involved in the detailed implementation of actual actions and situations. The respondent groups could not be separated as to the significance of the reflection, however, lack of time in the daily work was reproduced jointly by the respondents as an obstacle to the execution and prioritization of the reflection. The marked difference between the groups was found in the use of self-insight to develop their own leadership. One group actively described how they used self-insight and awareness about it, something that could not be identified in the other group that only reproduced presumed answers in the form of what should be done or overall answers about how the leader in general should act.

As more and more modern systems and products use built-in microcontrollers, hardware security becomes more important to protect against cyber-attacks. Internet of things devices, like Bluetooth devices, usually use an encryption algorithm to keep data safe from hackers. Advanced Encryption Standard (AES) is a commonly used encryption algorithm. AES itself is hard to break. However, it is possible to utilize the information leaking from a system during the execution of encryption, called side-channel, to recover the key or part of the key used by the encryption algorithm. This kind of attack is called a side-channel attack (SCA). In this study, two deep learning (DL) models are trained to attack the Bluetooth microcontroller unit Nordic nRF52 development kit equipped with an nRF52832 chip. The DL models are trained using the far-field electromagnetic emissions that the microcontroller unintentionally generates and transmits through the antenna while encrypting data. All encryptions are executed with a fixed key and random plaintext. The attack is conducted in two stages: the profiling and attack stages. In the profiling stage, where the attacker is assumed to have full system control, 100 000 traces holding encryption information are sampled and used to train the DL models to classify a sub-byte of the fixed key given a trace. In the attack stage, traces are captured in two different environments. The first is an entirely isolated environment, while the second adds a specific Wi-Fi access point and client connection that execute HTTP requests and responses in this isolated environment referred to as the system environment. Given traces obtained from one of the two attack environments, the performance of the trained models at classifying the correct sub-key is evaluated.  To summarize the results of this study, twelve SCAs are performed on six datasets captured in two different environments using two different DL models for each dataset. The correct key byte can be retrieved in three of these SCAs. All three successful attacks are made in an isolated environment without any interfering noise. The best performance is achieved with the multi-layer perceptron DL architecture, processing traces each composed of 10 averaged traces of the identical encryption, and the correct key-byte is recovered after 8198 traces.

The aim of this work is to enhance the detection and classification pipelines of an artificial intelligence (AI)-based decision support system (DSS) for diagnosing acanthamoeba keratitis (AK), a vision-threatening disease. The images used are taken with the in vivo confocal microscopy (IVCM) technique, a complementary tool for clinical assessment of the cornea that requires manual human analysis to support diagnosis. The DSS facilitates automated image analysis and currently aids in diagnosing AK. However, the accuracy of AK detection needs improvements in order to use it in clinical practice. To address this challenge, we utilize image brightness processing through multiscale retinex (MSR), and develop a custom-built image processing pipeline with deep learning model and rule-based strategies. The proposed pipeline replaces two deep learning models in original DSS, resulting in an overall accuracy improvement of 10.23% on average. Additionally, our improved pipeline not only enhances the original system’s ability to aid AK diagnosis, but also provides a versatile set of functions that can be used to create pipelines for detecting similar keratitis diseases.

DeepPaint is a semi-automated segmentation tool that utilises a U-net architecture to performbinary segmentation. To maximise the model’s performance and minimise user time, it isadvisable to apply Transfer Learning (TL) and reuse a model trained on a similar segmentationtask. However, due to the sensitivity of medical data and the unique properties of certainsegmentation tasks, TL is not feasible for some applications. In such circumstances, SelfSupervised Learning (SSL) emerges as the most viable option to minimise the time spent inDeepPaint by a user. Various pretext tasks, exploring both corruption segmentation and corruption restoration, usingsuperpixels and square patches, were designed and evaluated. With a limited number ofiterations in both the pretext and downstream tasks, significant improvements across fourdifferent datasets were observed. The results reveal that SSL models, particularly those pretrained on corruption segmentation tasks where square patches were corrupted, consistentlyoutperformed models without pre-training, with regards to a cumulative Dice SimilarityCoefficient (DSC). To examine whether a model could learn relevant features from a pretext task, Centred KernelAlignment (CKA) was used to measure the similarity of feature spaces across a model's layersbefore and after fine-tuning on the downstream task. Surprisingly, no significant positivecorrelation between downstream DSC and CKA was observed in the encoder, likely due to thelimited fine-tuning allowed. Furthermore, it was examined whether pre-training on the entiredataset, as opposed to only the training subset, yielded different downstream results. Asexpected, significantly higher DSC in the downstream task is more likely if the model hadaccess to all data during the pretext task. The differences in downstream segmentationperformance between models that accessed different data subsets during pre-training variedacross datasets.