Les systèmes d’information décisionnels dans le cloud Computing sont des solutions de plus en plus répandues. En effet, ces dernières offrent des capacités pour l’aide à la décision via l’élasticité des ressources pay-per-use du Cloud. Toutefois, les questions de sécurité des données demeurent une des principales préoccupations notamment lorsqu'il s’agit de traiter des données sensibles de l’entreprise. Beaucoup de questions de sécurité sont soulevées en terme de stockage, de protection, de disponibilité, d'intégrité, de sauvegarde et de récupération des données ainsi que des transferts des données dans un Cloud public. Les risques de sécurité peuvent provenir non seulement des fournisseurs de services de cloud computing mais aussi d’intrus malveillants. Les entrepôts de données dans les nuages devraient contenir des données sécurisées afin de permettre à la fois le traitement d'analyse en ligne hautement protégé et efficacement rafraîchi. Et ceci à plus faibles coûts de stockage et d'accès avec le modèle de paiement à la demande. Dans cette thèse, nous proposons deux nouvelles approches pour la sécurisation des entrepôts de données dans les nuages basées respectivement sur le partage vérifiable de clé secrète (bpVSS) et le partage vérifiable et flexible de clé secrète (fVSS). L’objectif du partage de clé cryptée et la distribution des données auprès de plusieurs fournisseurs du cloud permet de garantir la confidentialité et la disponibilité des données. bpVSS et fVSS abordent cinq lacunes des approches existantes traitant de partage de clés secrètes. Tout d'abord, ils permettent le traitement de l’analyse en ligne. Deuxièmement, ils garantissent l'intégrité des données à l'aide de deux signatures interne et externe. Troisièmement, ils aident les utilisateurs à minimiser le coût de l’entreposage du cloud en limitant le volume global de données cryptées. Sachant que fVSS fait la répartition des volumes des données cryptées en fonction des tarifs des fournisseurs. Quatrièmement, fVSS améliore la sécurité basée sur le partage de clé secrète en imposant une nouvelle contrainte : aucun groupe de fournisseurs de service ne peut contenir suffisamment de volume de données cryptées pour reconstruire ou casser le secret. Et cinquièmement, fVSS permet l'actualisation de l'entrepôt de données, même si certains fournisseurs de services sont défaillants. Pour évaluer l'efficacité de bpVSS et fVSS, nous étudions théoriquement les facteurs qui influent sur nos approches en matière de sécurité, de complexité et de coût financier dans le modèle de paiement à la demande. Nous validons également expérimentalement la pertinence de nos approches avec le Benchmark schéma en étoile afin de démontrer son efficacité par rapport aux méthodes existantes. / Cloud business intelligence is an increasingly popular solution to deliver decision support capabilities via elastic, pay-per-use resources. However, data security issues are one of the top concerns when dealing with sensitive data. Many security issues are raised by data storage in a public cloud, including data privacy, data availability, data integrity, data backup and recovery, and data transfer safety. Moreover, security risks may come from both cloud service providers and intruders, while cloud data warehouses should be both highly protected and effectively refreshed and analyzed through on-line analysis processing. Hence, users seek secure data warehouses at the lowest possible storage and access costs within the pay-as-you-go paradigm.In this thesis, we propose two novel approaches for securing cloud data warehouses by base-p verifiable secret sharing (bpVSS) and flexible verifiable secret sharing (fVSS), respectively. Secret sharing encrypts and distributes data over several cloud service providers, thus enforcing data privacy and availability. bpVSS and fVSS address five shortcomings in existing secret sharing-based approaches. First, they allow on-line analysis processing. Second, they enforce data integrity with the help of both inner and outer signatures. Third, they help users minimize the cost of cloud warehousing by limiting global share volume. Moreover, fVSS balances the load among service providers with respect to their pricing policies. Fourth, fVSS improves secret sharing security by imposing a new constraint: no cloud service provide group can hold enough shares to reconstruct or break the secret. Five, fVSS allows refreshing the data warehouse even when some service providers fail. To evaluate bpVSS' and fVSS' efficiency, we theoretically study the factors that impact our approaches with respect to security, complexity and monetary cost in the pay-as-you-go paradigm. Moreover, we also validate the relevance of our approaches experimentally with the Star Schema Benchmark and demonstrate its superiority to related, existing methods.

The use of Machine Learning (ML) in business has increased significantly over the past years. Creating high quality and robust models requires a lot of data, which is at times infeasible to obtain. As more people are becoming concerned about their data being misused, data privacy is increasingly strengthened. In 2018, the General Data Protection Regulation (GDPR), was announced within the EU. Models that use either sensitive or personal data to train need to obtain that data in accordance with the regulatory rules, such as GDPR. One other data related issue is that enterprises who wish to collaborate on model building face problems when it requires them to share their private corporate data [36, 38]. In this thesis we will investigate how one might overcome the issue of directly accessing private data when training ML models by employing Federated Learning (FL) [38]. The concept of FL is to allow several silos, i.e. separate parties, to train models with the same objective, using their local data and then with the learned model parameters create a central model. The objective of the central model is to obtain the information learned by the separate models, without ever accessing the raw data itself. This is achieved by averaging the separate models’ weights into the central model. FL thus facilitates opportunities to train a model on large amounts of data from several sources, without the need of having access to the data itself. If one can create a model with this methodology, that is not significantly worse than a model trained on the raw data, then positive effects such as strengthened data privacy, cross-enterprise collaboration and more could be attainable. In this work we have used a financial data set consisting of 25242 equity research reports, provided by Skandinaviska Enskilda Banken (SEB). Each report has a recommendation label, either Buy, Sell or Hold, making this a multi-class classification problem. To evaluate the feasibility of FL we fine-tune the pre-trained Transformer model AlbertForSequenceClassification [37] on the classification task. We create one baseline model using the entire data set and an FL model with different experimental settings, for which the data is distributed both uniformly and non-uniformly. The baseline model is used to benchmark the FL model. Our results indicate that the best FL setting only suffers a small reduction in performance. The baseline model achieves an accuracy of 83.5% compared to 82.8% for the best FL model setting. Further, we find that with an increased number of clients, the performance is worsened. We also found that our FL model was not sensitive to non-uniform data distributions. All in all, we show that FL results in slightly worse generalisation compared to the baseline model, while strongly improving on data privacy, as the central model never accesses the clients’ data. / Företags nyttjande av maskininlärning har de senaste åren ökat signifikant och för att kunna skapa högkvalitativa modeller krävs stora mängder data, vilket kan vara svårt att insamla. Parallellt med detta så ökar också den allmänna förståelsen för hur användandet av data missbrukas, vilket har lätt till ett ökat behov av starkare datasäkerhet. 2018 så trädde General Data Protection Regulation (GDPR) i kraft inom EU, vilken bland annat ställer krav på hur företag skall hantera persondata. Företag med maskininlärningsmodeller som på något sätt använder känslig eller personlig data behöver således ha fått tillgång till denna data i enlighet med de rådande lagar och regler som omfattar datahanteringen. Ytterligare ett datarelaterat problem är då företag önskar att skapa gemensamma maskininlärningsmodeller som skulle kräva att de delar deras bolagsdata [36, 38]. Denna uppsats kommer att undersöka hur Federerad Inlärning [38] kan användas för att skapa maskinlärningsmodeller som överkommer dessa datasäkerhetsrelaterade problem. Federerad Inlärning är en metod för att på ett decentraliserat vis träna maskininlärningsmodeller. Detta omfattar att låta flera aktörer träna en modell var. Varje enskild aktör tränar respektive modell på deras isolerade data och delar sedan endast modellens parametrar till en central modell. På detta vis kan varje enskild modell bidra till den gemensamma modellen utan att den gemensamma modellen någonsin haft tillgång till den faktiska datan. Givet att en modell, skapad med Federerad Inlärning kan uppnå liknande resultat som en modell tränad på rådata, så finns många positiva fördelar så som ökad datasäkerhet och ökade samarbeten mellan företag. Under arbetet har ett dataset, bestående av 25242 finansiella rapporter tillgängliggjort av Skandinaviska Ensilda Banken (SEB) använts. Varje enskild rapport innefattar en rekommendation, antingen Köp, Sälj eller Håll, vilket innebär att vi utför muliklass-klassificering. Med datan tränas den förtränade Transformermodellen AlbertForSequence- Classification [37] på att klassificera rapporterna. En Baseline-modell, vilken har tränats på all rådata och flera Federerade modellkonfigurationer skapades, där bland annat varierande fördelningen av data mellan aktörer från att vara jämnt fördelat till vara ojämnt fördelad. Resultaten visar att den bästa Federerade modellkonfigurationen endast presterar något sämre än Baseline-modellen. Baselinemodellen uppnådde en klassificeringssäkerhet på 83.5% medan den bästa Federerade modellen uppnådde 82.8%. Resultaten visar också att den Federerade modellen inte var känslig mot att variera fördelningen av datamängd mellan aktorerna, samt att med ett ökat antal aktörer så minskar klassificeringssäkerheten. Sammanfattningsvis så visar vi att Federerad Inlärning uppnår nästan lika goda resultat som Baseline-modellen, samtidigt så bidrar metoden till avsevärt bättre datasäkerhet då den centrala modellen aldrig har tillgång till rådata.

Business simulation games are used in educational institutions and various industries in the private and public sector to train students and employees to practice the principles of management and decision-making skills by providing a fail-safe environment and enabling them to reflect on their simulation results. These games are generally advanced multiuser environments where a user, or a group of users, have access to a virtual company for making business decisions. Some of these games are expensive and their licences are time limited; typically, such a licence is only valid during the duration of the course. In general, these games are not available for the public as part of informal instructional courses. In Canada, teaching informal courses to immigrants and refugees, which involve data-driven decision making, to prepare them for future challenges they might encounter as business owners, can be challenging; especially considering language barriers and non-business-related backgrounds obtained outside of Canada. Furthermore, based on their decision-making styles, cognitive limitations, and past experiences, people may have an inaccurate perception of the problem or challenge they face, this could lead to poor decision making of the team they are part of and, therefore, this could reflect in the effectiveness of an organization as a whole. The objective of this research is to enrich current teaching tools in decision-making processes in entrepreneurship courses for newcomers in Canada with a comprehensive and visual representation of operational business problems involved in Business Intelligence and data analytics. More specifically, we designed and developed a 3D Business Simulation Game with randomized scenarios using modern technologies, such as Unreal Engine as the game engine; Adobe Fuse for the character creation, Mixamo for animation of the character, and Substance Painter for textures and materials for the assets. The research was conducted with the participation of the students of the Business Creation and Project Management course at VIRCS (Victoria Immigrant and Refugee Centre Society) where we tested this game on each one of the five units of the course. After designing, developing, and testing the 3D business simulation game, we conducted a comprehensive evaluation to investigate whether the decisions students made while playing were correct or not. We also evaluated whether they felt that the challenges were easier to understand, both as a team and individually, when they used the 3D business simulation game compared to only the written description of the problems. The main results we obtained from our study are the following: After playing the business simulation game, students became more aware of the importance of making correct decisions in different business scenarios. They made sure that the whole team understood the problem, and they felt generally good about their understanding of the course content. We also noticed that when the animation was not part of the business simulation game, they seemed to be confused when following written instructions. This indicate that they depended on the animations for their decision-making. We believe that, in some ways, the course and the 3D business simulation game we created for this research were a great opportunity to observe students becoming more confident in their future in Canada as entrepreneurs. We observed that, once the game has been used, the students become more participatory in class, the discussion of the course material increases, and in general, the students seem to enjoy the course more. / Graduate

No description available.

Digital implants have become a new frontier for body hackers, technology enthusiasts and disruptive innovation developers, who seek to service this technology for themselves and to new users. This thesis has explored conditions for future user adoption of human body augmentation with digital implants. The conditions explored were mainly self-beneficial health optimization through technology, self-quantification or convenience scenarios. Applying Diffusion Of Innovation theory, Value-based Acceptance Model and research through design methods were used. The process consisted of quantitative and qualitative data gathering and analysis, using interviews, surveys and iterative prototyping with evaluation. The results show mixed user attitude towards implant usage, mainly depending on users' need for added benefits, whether the user is a technology enthusiast actively using technology for self-beneficial gain or a casual everyday consumer of technology. Certain conditions could affect adoption of implants into mainstream usage, mainly data privacy, regulation, convenience, self-quantification or health management. In order for implants to succeed as a mainstream technology, there needs to be proper secure infrastructure, easy installation and coordinated services that offer individual benefits of health or convenience, with a high consumer confidence in supported services, installation / removal and devices. Several companies are working on offering such a service, in order to evaluate such a proposition, iterative prototypes were created to evaluate a health management scenario as a streamlined consumer service, using a service design blueprint and a related interactive smartphone application prototype.

No description available.

<p>Data Governance and Data Privacy are critical aspects of organizational management that are widely utilized across all organizational scales. However, this research focused specifically on the significance of Data Governance and Data Privacy in Small and Medium Enterprises (SMEs). While the importance of maintaining these systems is paramount across all organizations, the challenges faced by SMEs in maintaining these systems are greater due to their limited resources. These challenges include potential errors such as data leaks, use of corrupted data, or insufficient data, as well as the difficulty in identifying clear roles and responsibilities regarding data handling. To address these challenges, this research investigated the impact of utilizing Lean Six Sigma (LSS) tools and practices to overcome the anticipated gaps and challenges in SMEs. The qualitative methodology utilized is a grounded theory design, chosen due to the limited understanding of the best LSS practices for achieving data governance and data privacy in SMEs and how LSS can improve the adoption of data governance concerning privacy in SMEs. Data were collected using semi-structured interview questions that were reviewed by an expert panel and pilot tested. The sampling method included purposive, snowballing, and theoretical sampling, resulting in 20 participants being selected for interviews. Open, axial, and selective coding were performed, resulting in the development of a grounded theory. The obtained data were imported into NVivo, a qualitative analysis software program, to compare responses, categorize them into themes and groups, and develop a conceptual framework for Data Governance and Data Privacy. An iterative data collection and analysis approach was conducted to ensure that all aspects were considered. The applied grounded theory resulted in retrieving the themes used to generate a theory from the participants’ descriptions of LSS, SMEs, data governance, and data privacy. Finally, ISM technique has been applied to identify the relationships between the concepts and factors resulted from the grounded theory. It helps arranging the levels the criteria, drawing the relationships in a flowchart, and providing valuable insights to the researcher. </p>

This study focuses on the impact of Cloud-Based Shadow IT on data privacy in the tech sector of Sweden. It explores the use of unapproved applications by employees without the knowledge and control of the IT department. The objective is to understand how Cloud-Based Shadow IT affects employees' compliance with cloud services and to examine the understanding of IT managers regarding this phenomenon. The research problem addresses the challenges faced in ensuring compliance with regulations and effective utilization of cloud technology. By identifying the strengths, weaknesses, possibilities, and risks associated with Cloud-Based Shadow IT, this study aimed to provide insights for companies and IT managers in making informed decisions. It explores the relationship between Shadow IT and cloud services and investigates employees' and IT managers' adherence to and understanding of these issues.    The findings indicate that employees have varying levels of understanding, with limited knowledge of approved cloud services. Managers prioritize security concerns, including data compliance and ownership, but lack strategies to address knowledge gaps. The use of Cloud-Based Shadow IT has both positive and negative consequences, including increased productivity and collaboration but also data loss and non-compliance risks. Factors such as education and awareness of security risks are important for employees to understand and comply with policies. Overall, the study highlights the need for continuous education and awareness programs to improve understanding and decision-making regarding cloud services and Shadow IT.

Sensitive data leaking from a system can have tremendous negative consequences, such as discrimination, social stigma, and fraudulent economic consequences for those whose data has been leaked. Therefore, it’s of utmost importance that sensitive data is not leaked from a system. This thesis investigated different methods to prevent sensitive patient data from leaking in a machine learning system. Various methods have been investigated and evaluated based on previous research; the methods used in this thesis are a large language model (LLM) for code analysis and a membership inference attack on models to test their privacy level. The LLM code analysis results show that the Llama 3 (an LLM) model had an accuracy of 90% in identifying malicious code that attempts to steal sensitive patient data. The model analysis can evaluate and determine membership inference of sensitive patient data used for training in machine learning models, which is essential for determining data leakage a machine learning model can pose in machine learning systems. Further studies in increasing the deterministic and formatting of the LLM‘s responses must be investigated to ensure the robustness of the security system that utilizes LLMs before it can be deployed in a production environment. Further studies of the model analysis can apply a wider variety of evaluations, such as increased size of machine learning model types and increased range of attack testing types of machine learning models, which can be implemented into machine learning systems. / Känsliga data som läcker från ett system kan ha enorma negativa konsekvenser, såsom diskriminering, social stigmatisering och negativa ekonomiska konsekvenser för dem vars data har läckt ut. Därför är det av yttersta vikt att känsliga data inte läcker från ett system. Denna avhandling undersökte olika metoder för att förhindra att känsliga patientdata läcker ut ur ett maskininlärningssystem. Olika metoder har undersökts och utvärderats baserat på tidigare forskning; metoderna som användes i denna avhandling är en stor språkmodell (LLM) för kodanalys och en medlemskapsinfiltrationsattack på maskininlärnings (ML) modeller för att testa modellernas integritetsnivå. Kodanalysresultaten från LLM visar att modellen Llama 3 hade en noggrannhet på 90% i att identifiera skadlig kod som försöker stjäla känsliga patientdata. Modellanalysen kan utvärdera och bestämma medlemskap av känsliga patientdata som används för träning i maskininlärningsmodeller, vilket är avgörande för att bestämma den dataläckage som en maskininlärningsmodell kan exponera. Ytterligare studier för att öka determinismen och formateringen av LLM:s svar måste undersökas för att säkerställa robustheten i säkerhetssystemet som använder LLM:er innan det kan driftsättas i en produktionsmiljö. Vidare studier av modellanalysen kan tillämpa ytterligare bredd av utvärderingar, såsom ökad storlek på maskininlärningsmodelltyper och ökat utbud av attacktesttyper av maskininlärningsmodeller som kan implementeras i maskininlärningssystem.

Lite-Agro studies applications of TinyML in pear (Pyrus communis) tree disease identification and explores hardware implementations with an ESP32 microcontroller. The study works with the DiaMOS Pear Dataset to learn through image analysis whether the leaf is healthy or not, and classifies it according to curl, healthy, spot or slug categories. The system is designed as a low cost and light-duty computing detection edge solution that compares models such as InceptionV3, XceptionV3, EfficientNetB0, and MobileNetV2. This work also researches integration with federated learning frameworks and provides an introduction to federated averaging algorithms.