Högskolebiblioteket i Halmstad Publikationer
Not a member yet
17134 research outputs found
Sort by
Android Malware Classification Using Explainable AI
No full textAndroid malware is a growing cybersecurity concern, with attackers constantly developing new techniques to bypass traditional detection mechanisms. Artificial Intelligence (AI) and Machine Learning (ML)based detection techniques have shown promising results in identifying malicious applications. However, their lack of interpretability poses a significant challenge. Many existing models function as " black boxes," making it difficult for security analysts to understand why an application is classified as malicious or benign. This lack of transparency limits trust, hinders forensic investigations, and reduces the effectiveness of cybersecurity defences. Our research explores the use of Explainable AI (XAI) techniques in Android malware classification, addressing the topic of interpretability in these "black box" models. We employ the CICMalDroid 2020 dataset to evaluate a range of machine learning models, including Decision Tree, Random Forest, XGBoost, Multi-Layer Perceptron (MLP), LightGBM, and Support Vector Machines (SVM), for Android malware detection. To enhance the transparency and interpretability of these complex models, we apply post-hoc explainable AI techniques, specifically SHAP (Shapley Additive Explanations) and LIME (Local Interpretable Model-Agnostic Explanations), enabling detailed analysis of feature contributions to classification outcomes. By integrating these XAI techniques, our approach aims to improve detection accuracy and offer insight into which features contribute more to malware detection and classification of Android applications, thereby helping security analysts make informed decisions.
UX Designer’s Perception of Trust in AI-Generated Design Recommendation
No full textAs AI becomes more integrated into UX design tools, building trust in AI-generated output is important for successful adoption. This thesis investigates how UX designers perceive and build trust in AI-assisted design tools including Figma AI, ChatGPT and Adobe Sense. Using a qualitative approach, semi-structured interviews with professional fissional were conducted to investigate their interactions with AI-generated content. The findings build on the Hoff and Bashir Trust Model by identifying six trust influencing themes: initial expectations, trust criteria, task-specific usage, human oversight, consistency and ethical concerns. These findings provide practical insights into how designers evaluate, modify, or reject AI proposals, highlighting the significance of transparency, control and inclusion in AI design. The findings contribute to a better understanding of trust in human-AI collaboration and can help developers create more trustworthy, user-centered AI tools for design workflows
The impact of urban planning and real estate development on property value and profitability
No full textAbstract This bachelor's thesis aims to investigate how urban planning and development affect profitability and property value, as well as which measures and improvements property companies can invest in to create increased value. The study is based on an abductive approach and employs a qualitative method, with empirical data gathered through semi-structured interviews with professionals in the real estate industry. The theoretical framework addresses key concepts such as property value, profitability, placemaking, urban strategies, and property development. The results show that urban planning and placemaking have a significant impact on both profitability and property value. Key contributing factors include improved accessibility, enhanced attractiveness, and increased foot traffic. Property owners emphasize that investments in safety, aesthetics, green spaces, and adaptation to the identity of the location contribute to long-term value creation. Collaboration with municipalities and other stakeholders appears to be a crucial factor, as does a holistic perspective that considers both economic and social values. The study concludes that strategic property development, with a clear connection to the site's character and potential, is essential for achieving increased profitability. Creating value requires proactive and long-term efforts where property companies invest in both the physical environment and the relationships between stakeholders in the area
A Thermal-Fluid Approach to Sauna Room Modelling and Analysis : CFD-Based Investigation of Heat Transfer and Energy Efficiency in Modern Sauna Rooms
No full textThis thesis presents a comprehensive thermal-fluid analysis of sauna room configurations using Computational Fluid Dynamics (CFD) to evaluate heat transfer, airflow patterns, and energy efficiency. Two room layouts are compared: a traditional Glass Door (GD) setup and a modern Glass Front (GF) design, with a focus on the thermal implications of using glass versus wood paneling. The study employs the realizable k-epsilon turbulence model with enhanced wall treatment to simulate heat transfer by buoyancy-driven natural convection within the sauna environment and includes radiation modelling via the Discrete Ordinates (DO) method. First-order analyses based on the heat diffusion equation were conducted to support the simulation results and guide boundary condition setup. Validation was performed against experimental data from a real-world sauna lab, showing good agreement with an average deviation of 7%. The results reveal that the GF configuration increases energy demand by 26% compared to the GD setup, primarily due to the higher emissivity and lack of insulation in glass surfaces, despite minimal differences in flow behavior and thermal stratification. The findings suggest that design choices significantly impact operational efficiency, with implications for cost, environmental sustainability, and user comfort
Quantitative measurements and influence assessment on chemical surface cleanliness
No full textThis thesis investigates the relationship between chemical surface contamination andadhesion performance in industrial applications, focusing on insulation tape used inScania’s battery module production. Two manufacturing challenges are addressed: poorlaser weld quality and tape adhesion failures, both stemming from surface contaminationby oils and greases.Due to equipment limitations and time constraints, the study prioritizes the developmentof methods for contamination measurement and evaluating the effects on tape adhesion.Interferometry, fluorescence spectroscopy, and gravimetric analysis were applied toquantify surface cleanliness, and a custom-built tape peel test rig was used to simulatereal-world adhesion performance.Experiments demonstrated a measurable correlation between increased surfacecontamination and reduced adhesion force. Fluorescence measurements proved effectivein detecting relative contamination levels, though accurate quantification depended onsubstrate type and oil properties. A strong link between fluorescence intensity andadhesion strength was observed for uncoated steel plates, highlighting the technique’spotential for quality control applications.The results support Scania’s need to define contamination thresholds and establishprocess reliability, even when the exact contaminant is unknown.
Development of Fuzzy Logic Controllers for Non-Linear Industrial Processes
No full text This thesis work outlines the design and simulation of a fuzzy logic control system fortemperature or pressure regulation in automated industrial processes. Although conventional Proportional-Integral-Derivative (PID) controllers are ubiquitous, they lose their optimal performance in nonlinear and dynamic processes. Fuzzy logic controllers (FLCs), on the other hand, provide an intelligent and adaptive solution by simulating human decision-making using rule-based inference and linguistic variables. A simulation model was developed usingMATLAB/Simulink to compare the performance of the FLC with that of a conventional PID controller in regulating the temperature or pressure in a chemical reactor The fuzzy controller used Gaussian membership functions and a predefined rule base, allowing real-time decision-making based on both the error and rate of change. Simulation results verified that FLCs were superior to PID controllers in stability, response time, adaptability, and robustness underdynamic operating conditions. Additionally, the study highlighted the modularity, computational efficiency, and scalability of FLCs, thereby justifying their practical applicability in industrial automation. The project concludes by recommending the integration of FLCs into control systems to promote efficiency and flexibility, while outlining potential areas of research in hardware implementation and hybrid intelligent systems aimed at improving process contro
Finansiell kunskaps inverkan på kapitalstruktur : En kvalitativ studie om beslutsfattares finansiella kunskap
No full text Title: Financial literacy’s influence on the capital structure. A qualitative study on decision makers financial literacy's Subject: Bachelor’s thesis in business administration, 15 credits Keywords: Financial literacy, Capital structure, Real estate companies, Small and medium enterprises (SME) Purpose: The purpose of this study is to create an understanding of how the decisions makers financial literacy has an impact on the capital structure in small real estate companies Method: The study has been conducted with a abductive approach, which involves an interplay between theory and empirics, allowing the theory to be adjusted based on insights that surfaced from the empirical data. Results: This study presents four main findings. The first finding is that financial literacy exists as an asset within the companies, but not necessarily with the decision makers themselves. The second finding shows that the level of financial literacy influences how the company approaches financing. Applied knowledge contributed to more opportunities and alternative choices, while basic knowledge resulted in more traditional options such as bank loans. Basic knowledge led companies to follow the pecking order theory, while deviations from the theory were linked to more comprehensive knowledge. The third finding highlights that contextual knowledge can be as important as financial literacy. Finally, the study shows that financial literacy reduces information asymmetry and allows better communication with external actors, which in turn leads to higher debt levels and impacts the capital structure
Agro-biomass to Biogas : Drivers and barriers in Sweden
No full textRenewable alternatives like biogas are becoming more and more necessary as the world's energy needs continue to rise and climate worries grow. A concerted effort is underway in Sweden to use agro-biomass to increase biogas generation. The socioeconomic and environmental factors, obstacles, and tactics associated with the sustainable mobilization of agro-biomass in Sweden to increase biogas production are examined in this study. Qualitative data were gathered from surveys with 14 Swedish farmers and stakeholders, complemented by a stakeholder interview. Thematic analysis was used to organize the findings into categories of benefits, barriers, and strategies. The results indicate that sustainable farming practices and potential feedstocks, such as grassleys, straw from cereals and oilseeds, the incorporation of catch or cover crops, intermediate crops, and the adoption of sequential cropping systems, can significantly enhance biomass production. The study identified several motivating factors for stakeholders, with economic incentives and environmental considerations taking precedence, followed by government policies, energy efficiency, and biofertilizer digestate as an efficient manure. On the other hand, the primary obstacles revealed include environmental concerns such as land use, habitat loss, soil quality degradation, high costs, and the lack of adequate policy regulations. Accordingly, this study comes to the conclusion that although agro-biomass has a great deal of potential to increase Sweden's biogas output, realizing this potential would require the development of unified policies, efficient subsidies, and better stakeholder cooperation. As Sweden moves forward with its transition to renewable energy, this study emphasizes the crucial connection between theoretical promise and real-world difficulties, emphasizing the need for more precise regulations, financial incentives, and specialized assistance for farms
Get Out of Jail Free Prompt : How Jailbreaking Compromises Safety in Large Language Models
No full textThe increasing popularity of large language models (LLM) has raised important concerns regardingtheir security and the implications of their misuse. This thesis evaluates the vulnerability of LLMs toknown jailbreaking techniques and the potential harm caused by successful attacks. Using a structuredscoring framework, three popular LLMs (GPT-4o, DeepSeek-V3, and Mistral’s Le Chat) were testedagainst five jailbreaking techniques. All models were susceptible to varying degrees, with DeepSeek-V3 being the most vulnerable, particularly to role-playing. Although most jailbroken responses wereconsidered Low Harm, a few were highly actionable with weak ethical disclaimers, underscoring real-world risks. The thesis emphasizes the importance of prioritizing improved pattern recognition, strictercontextual adherence, and responsible access controls for sensitive capabilities, especially when itcomes to cybersecurity-related content. Until these issues are addressed, jailbreaking will remain athreat requiring proactive and adaptive mitigation strategies