[dipl] Perustieteiden korkeakoulu / SCI

Permanent URI for this collection

Browse

Recent Submissions

Now showing 1 - 20 of 4874
  • Item
    Attention-based method to predict drug-target interactions across seven protein classes
    (2024-01-22) Schulman, Aron; Tanoli, Ziaurrehman; Perustieteiden korkeakoulu; Rousu, Juho
    Most approved drugs bind with proteins to modulate their activity for treating a diverse range of diseases. Unfortunately, drug development is a long and costly process. Computational methods seek to accelerate drug discovery by predicting drug-target interactions, thus facilitating compound screening and drug repurposing. This thesis presents a deep learning approach for predicting interactions between compounds and proteins categorized into seven classes. The models utilize self-attention found in transformer neural networks to learn continuous interaction values from multimodal compound-protein representations. The models were evaluated in three test settings of increasing difficulty. Most models showed competitive predictive capabilities in the two easier settings, with the most difficult test rendering them ineffective. In particular, the kinase model demonstrated state-of-the-art performance in the bioactivity imputation task when compared against other methods, while leaving room for improvement in the new compound scenario. Furthermore, conformal predictors with uncertainty estimates displayed equivalent performance to contemporary methods and provided directions for future research.
  • Item
    Usability evaluation of a robotic-guided multi-locus transcranial magnetic stimulation (mTMS) system
    (2024-01-22) Kauppi, Henrik; Souza, Victor; Perustieteiden korkeakoulu; Viitanen, Johanna
    As medical systems grow more complex, an increasing emphasis on evaluating and improving the usability of these systems is required to prevent potentially dangerous user errors and to fix existing usability problems. However, usability evaluation of complex and domain-specific systems presents unique challenges, such as lacking domain knowledge, and choosing the optimal evaluation methods. Careful planning is required to mitigate these challenges and find usability problems that are the most relevant to the users. The two goals of this thesis include: 1) finding usability problems in the robotic-guided mTMS system in a research context and creating improvement suggestions and 2) Analyzing the difference in usability between manual operation of the system and robotic operation of the system based on the results of the usability evaluation. The usability evaluation was conducted using four evaluation methods: heuristic evaluation, user testing, semi-structured interview, and System Usability Scale (SUS) questionnaire. Additionally, I also facilitated a workshop to validate the improvement suggestions and prototypes. The usability evaluation recognized a total of 36 usability problems with the system, with efficiency and error prevention being the two most problematic usability attributes. The findings showed that although the system is currently usable by expert users for research purposes, there are still lots of usability problems, some of which are severe. The findings also showed that the system was easier to use with robotic operation when compared to manual operation.
  • Item
    Organizing funding in Finnish hardware startups
    (2024-01-24) Nieminen, Mikael; Jääskeläinen, Mikko; Perustieteiden korkeakoulu; Jääskeläinen, Mikko
    Access to sufficient funding is critical for the success of any startup, but this is a particularly relevant topic for hardware startups that integrate electronic, mechanical, and software components into complex products. Hardware development is difficult, complex, expensive, and slow in nature, which requires special considerations in funding and can cause funding related challenges, especially during the early company stage. This study set out to investigate the most suitable ways to finance hardware startups during the product development phase, and how the characteristics of hardware development affect the funding practices. Geographical scope of the empirical work was placed in Finland by conducting in-depth interviews for 12 Finnish hardware startups. These case companies were analyzed with diverse methods to build theory and provide recommendations for hardware startup funding with a tight connection to the empirical evidence. The main results of the study were the following. First, funding was a major challenge for most of the case companies, and positioning closer to software can make it easier. Second, Business Finland and angel investors were the most common early investors, while most of the case companies made market entry without VCs. Third, three different types of funding and growth paths for hardware startups were identified within the case companies: 1) fast growth with large and early VC investments, 2) slower and organic revenue-based growth, and 3) generic startup growth path. These paths are introduced in detail to help similar companies to position themselves in relation to factors that guide the most feasible financing strategies. The key factors guiding path selection include team experience and competence, aspirations and ambition level of the company, the availability of initial technologies and other assets, product type, and business model. Finally, two key factors providing advantage for hardware startups are high experience of the founding team and availability of initial key assets. These are also factors that favour the selection of funding and growth path numbers 1 and 2. A key implication of this study is that hardware is a difficult startup category to obtain funding for, and the development timelines, capital requirements, and general difficulty of development must be carefully considered in financing. Another implication is that most investors do not look at hardware startups through a very technical lens or categorize them completely apart from other startups. Hence, the general startup funding principles apply for hardware startups too, and they should focus on providing the investors a good business case instead of focusing too much on product details.
  • Item
    Multi-Agent Reinforcement Learning for Littoral Naval Warfare
    (2024-01-22) Vasankari, Lauri; Heikkonen, Jukka; Perustieteiden korkeakoulu; Pajarinen, Jouni
    Littoral warfare refers to naval combat that occurs near coastlines or in seas with more constraints than the vast open oceans. Although the fundamental principles of naval warfare are consistent in littoral zones, this environment introduces unique challenges and demands specific tactics. Tactical doctrines must recognize the strengths and limitations of deploying forces in these areas, necessitating careful planning and decision-making. Decision-making in such scenarios is complex, requiring a balance between timeliness and thoroughness, often under conditions of limited resources and significant uncertainty. Machine learning has the potential to enhance decision-making in this context by predicting possible outcomes, detecting patterns, and determining optimal actions. Naval conflicts have been infrequent throughout history, leading to a scarcity of data for applying data-intensive machine learning techniques. Naval forces consist of various units, and naval warfare is inherently competitive, involving two or more opposing sides. This environment is represented as a Partially Observable Stochastic Game (POSG), featuring two teams managed by decision-making agents. In POSG, the game is modeled through a framework of agents, states, actions, observations, and the probabilities of transitioning between states based on these actions. Multi-Agent Reinforcement Learning (MARL) algorithms are employed to devise policies that approximate solutions to POSG, aiming to address the challenges posed by this setting. The study tests two MARL algorithms, Double Deep Q-Networks and Proximal Policy Optimization, within this multi-agent context. The findings suggest that despite complexities and the changing nature of the environment (non-stationarity), MARL approaches can generate and analyze different tactical options. This assists decision-makers by either reinforcing established tactical doctrines or offering new solutions, thereby enriching tactical planning and execution.
  • Item
    Development of Usability Guidelines for Construction Industry Internal Web Applications
    (2024-01-22) Keinänen, Markus; Panjwani, Mehdi; Perustieteiden korkeakoulu; Kujala, Sari
    This study was conducted at a Finnish construction consulting company. The company maintains several services developed by the internal software engineering team. The study was motivated by usability issues stemming from lack of usability knowledge. To remedy the situation, guidelines were formed for the context of internal web use related to construction consulting. The study uses the design science research approach to arrive at a research-based solution. The usability guidelines were designed by modifying a baseline set of guidelines according to the usability issues discovered. To gather contextual usability data, interviews were conducted among developers (N=3) and a survey was conducted with the users of the internal web applications (N=17). Personal judgement was discovered to be the most common usability evaluation method among the developers. Lack of intuitiveness and errors were discovered to be biggest usability problems for users, and lack of intuitiveness and inconsistency for the developers. When analysing each usability issue to determine the need for new guidelines, pre-existing guidelines were found to address most specific usability issues for both developers and users. New guidelines were added for unaddressed issues. Guidelines were not sufficient for addressing all discovered problems, such as lack of resources and intuitiveness. The result of the design process is an interactive web-based service, which lists guidelines alongside helpful illustrations to support the developers in usable design. At the end of the study, the service was evaluated by the developers through a survey (N=3) and found to be helpful in producing services with better usability. When developing usability guidelines for internal web apps, choosing specific pre-existing guidelines in the generic web context and modifying them to address context-specific usability issues appears to be a good approach.
  • Item
    The Effects of Environmental Factors on User Experience in Virtual Reality Intangible Cultural Heritage Experience
    (2024-01-22) Jia, Yingqi; Lu, Zhicong; Perustieteiden korkeakoulu; Hämäläinen, Perttu
    Virtual Reality (VR) has a unique capacity to create a sense of presence and immerse users in alternative worlds, narratives, and cultures. For traditional handicrafts (TH) in the category of intangible cultural heritage (ICH), VR has the ability to convey to the public the process and cultural atmosphere of creating handicrafts that are difficult to experience in real life. In order to increase public awareness of TH, we created a VR experience about Chinese Miao Batik, a low-popularity TH. We compare two VR experiences in a user study: one in daily production sites of Miao Batik craftsmen, and the other in a museum workshop. Our results suggest that i) VR can elicit cultural engagement and return visit interest, and ii) a VR experience depicting the realistic production environment with ethnocultural characteristics inspires higher user engagement and sense of presence. This study provides a valuable reference for designing ICH-related VR experiences.
  • Item
    An Augmented Reality Assisted Indoor Navigation System
    (2024-01-22) Soltani, Javad; Ylä-Jääski, Antti; Perustieteiden korkeakoulu; Ylä-Jääski, Antti
    Augmented reality-based location services address the demand for immediate indoor navigation, transforming individuals' perspectives and engagement with their surroundings in real-time. This thesis aims to investigate the technologies and applications of location-based services and develop a prototype for indoor navigation that implements augmented reality to assist users in locating themselves within a building using their smartphones. Indoor positioning technologies can be classified into radio frequency-based and non-radio frequency-based systems. Radio frequency signals are common technologies used for indoor positioning because of the widespread presence of wireless communication systems and their fast improvement. Non-radio frequency-based systems are computer vision, inertial measurement units, visible and non-visible light communication, ultrasonic signals, and magnetic fields. The precision levels vary from less than a centimeter to a few meters, and the infrastructure demands of these systems differ significantly, ranging from almost none to highly specialized and expensive equipment. Location-based services can be applied in various domains, such as guiding users through buildings, providing contextual information in real-time, games that leverage knowledge of physical surroundings, and healthcare services. In domains where smartphones are applicable, computer vision algorithms play a crucial role by processing the camera feed of devices. Consequently, it also enables augmented reality features, which blur the boundaries between developed applications and their surrendering physical world. The implemented augmented reality-assisted Unity application allows for seamless navigation within buildings. The application uses image processing and inertial sensors to locate users and provide guidance through 2D map checkpoints and 3D augmented markers. The results of experiments show that the prototype maintains consistency in tracking paths during navigation with an average positioning accuracy between 2 and 4 meters. The application makes Indoor navigation feasible, but there is room to enhance its performance.
  • Item
    Investigating different characterisation methods for protein-cellulose composites
    (2024-01-23) Pascale, Alexander; Tersteegen, Jennifer; Perustieteiden korkeakoulu; Linder, Markus
    The surge in sustainable materials development is a direct response to the demand for environmentally friendly alternatives. This study emphasises the relevance of drawing inspiration from the natural environment to discover new components for sustainable solutions. Biocomposites, marked by significant progress in diverse raw materials, processing techniques, and applications, present promising alternatives to traditional composite materials. The aim is to reduce the carbon footprint and environmental strain, with numerous studies investigating the enhancement of mechanical performance. This study investigates the characterisation of delignified cellulose samples infiltrated with Bovine Serum Albumin (BSA) protein using various methods due to inconsistent results with recombinant spider silk and cellulose. The analysis includes assessing protein infiltration depth in bulk cellulose and conducting mechanical tests on lap shear samples and cellulose stripes with different infiltration techniques. The impact of varied BSA concentrations on mechanical properties is explored. Results reveal that BSA concentration influences adhesiveness and mechanical performance for lap shear samples. Among tested concentrations, 80 mg/ml BSA displays high lap shear strength. However, the BSA 80 mg/ml shows high deviation on the test series. Therefore, further set ups were investigated. The aim is to receive a better understanding of the mechanical properties and the adhesive abilities of the BSA. Furthermore, using the knowledge gained from the BSA samples for other proteins, e.g. spider silk protein, and to find the best way of infiltrating the protein towards delignified wood. This study sheds light on cellulose-protein interactions and their potential applications, requiring further exploration for optimised outcomes.
  • Item
    Adversarial Robotic Cloth Manipulation
    (2024-01-22) Laitinen, Sanna; Blanco, David; Luck, Kevin; Perustieteiden korkeakoulu; Kyrki, Ville
    Robotic cloth manipulation is still a largely unresolved problem, even though many of our day-to-day tasks involve clothes. Generative adversarial methods have been applied for generating realistic and difficult tasks for rigid object manipulation algorithms. Continuing from this, the main contribution of this thesis is a novel algorithm for generating adversarial clothes. In addition, we propose a simple method for estimating the success of cloth unfolding for complex meshes during simulation time only based on images. The novel adversarial clothes generation algorithm was tested by training the FlingBot model with the generated clothes. Then we evaluated the model performance on four novel dynamic clothes unfolding data sets, consisting of skirts, trousers, shirts, and socks. To our knowledge such a variety of items has not been previously tested on dynamic cloth manipulation tasks. Our adversarially trained model performed equally well compared with a model trained on easy tasks and outperformed a model trained with domain randomized clothes.
  • Item
    On strategic dissonance and renewal: Navigating through rapid environmental changes in a Finnish cooperative bank
    (2024-01-24) Simola, Henri; Lähteenmäki, Ilkka; Perustieteiden korkeakoulu; Vuori, Timo
    This master’s thesis examines strategic dissonance—the acute misalignment between an organizations strategic intent and actions, and between basis of competition and distinct competence—in periods of rapid environmental change, focusing particularly on the origins, consequences, and management of dissonance. The study proposes that the onset of strategic dissonance triggers a process of strategic renewal in which organizations attempt to build new external and internal fit to respond to changes in their external environment while managing associated structural and cognitive conflicts. The research is conducted as a single case study in a Finnish cooperative bank, utilizing the Gioia methodology. The empirical data primarily consists of interviews with the company’s top management, business unit managers, and board members. The thesis presents a theoretical process model, which illustrates the combined effect of rapid environmental change, cognitive division, and structural enablers in the formation of strategic dissonance. Furthermore, the model highlights how dissonance manifests as an organizational and cognitive tug-of-war between established practices and emerging paradigms. The management of this conflict determines the quality of the new competences, capabilities, and alignment of resources in relation to changed environmental requirements and de- mands. If inadequately resolved or prolonged, it may lead to a vicious cycle of recurring dissonance. In addition, the thesis explores three management approaches to strategic recognition, i.e., estab- lishing new strategic intent and direction thereby charting a path towards renewal: authoritative, consensus-seeking, and balanced. These approaches have varying impacts on the organization’s resulting new fit and can either lead to recurring dissonance or contribute towards effective renewal. This master’s thesis integrates various, closely related research areas from literature on strategic dissonance, renewal, and managerial and organizational cognition. The proposed framework underscores the comprehensive nature of strategic change encompassing context, content, and process, enriching theory by mapping theoretical similarities and thereby providing an integrative perspective. Furthermore, the research aids practitioners in avoiding potential pitfalls of dissonance by recognizing underlying predisposing factors, and pursuing effective renewal by identifying potential pathways for renewal leadership.
  • Item
    Computer vision-assisted safety alert system for industrial settings in private 5G networks
    (2024-01-22) Chowdhury, Suha; Ahmad, Iftikhar; Perustieteiden korkeakoulu; Jung, Alexander
    High-risk activities and complex machinery create an ever-present potential for injuries and accidents in the industrial landscape. While significant advancements have taken place in computer vision applications over the last few decades, workplace injury rates in the manufacturing sector remain relatively high. 5G technology and computer vision techniques have shown great promise in improving working conditions through wireless connectivity and real-time object detection and tracking. However, researchers have not yet extensively explored their combined potential. This thesis presents a novel computer vision-assisted 5G alert system that addresses safety-critical scenarios in the manufacturing sector. Four safety-critical scenarios were developed and designed: restricted static zone entries, vehicle-to-vehicle potential collisions, human-to-vehicle potential collisions, and potential collisions between humans and unidentified mobile objects. The research included implementing two frameworks within the safety alert system: supervised and semi-supervised. The supervised framework involved multi-object detection and tracking of annotated objects in safety-critical scenarios, while the semi-supervised framework utilized motion detectors and zero-shot object segmentation for detecting and segmenting unidentified mobile objects. In the case of the semi-supervised framework, the performance of two motion detectors was evaluated. Furthermore, a novel two-stage approach was introduced for detecting and segmenting unlabeled objects. The study revealed that raw bounding box information was not a suitable metric for generating appropriately sized alert beams for sporadically rotating objects. In the case of the supervised framework, alerts received by a single user were examined, with the residence time in each restricted zone being quantified and documented. The findings highlighted the effective and timely response of the safety alert system in addressing safety concerns. The proposed work provides technical requirements for implementing the respective frameworks and offers insights into future research to deploy a computer vision-assisted 5G safety alert system.
  • Item
    Digital advantages in the platform economy
    (2024-01-24) Viheriävaara, Aarne; Karhu, Kimmo; Perustieteiden korkeakoulu; Karhu, Kimmo
    The advent and proliferation of digital platforms have significantly transformed traditional business models and consumer interactions. This thesis investigates the pivotal digital advantages that underpin the success of such platforms within the current economic landscape. Through an analysis of case data from diverse digital platforms, this research identifies and categorises the distinct digital advantages that these platforms leverage for transactional, creative, informational, productivity, network effects and strategic benefits. Additionally, the enabling digital mechanisms behind the digital advantages are uncovered, categorised, and mapped to each advantage, resulting in distinct combinations of the mechanisms that expose the diverse digital characteristics of digital platforms. The findings of this thesis contribute to the existing body of literature by offering a nuanced categorisation of digital advantages, and the digital mechanism enabling them, thereby addressing gaps in the literature regarding the characterisation and categorisation of digital platforms.
  • Item
    Improving Software Deployment for Mobile Machinery: A Case Study and Proof of Concept Implementation
    (2024-01-22) Mäkinen, Jaakko; Hannelius, Tom; Perustieteiden korkeakoulu; Lassenius, Casper
    To support agile and DevOps practices, continuous integration, delivery, and deployment have emerged as methods to provide a fast feedback loop, improved software quality, and rapid delivery of value to stakeholders. The rise of IoT and increasing reliance on software in industrial contexts has evoked the need to adopt continuous practices in these domains to accelerate R&D and maintain a competitive edge. A gap in research was identified, concerning the adoption of continuous deployment practices in the context of mobile machinery. This study assesses the current state of software deployment in industrial domains, using design science research for information systems. A small-scale interview-based case study was conducted to assess the current state of software deployment in industrial domains. The findings indicate that deployment and configuration activities currently require a lot of manual work, which is costly and time-consuming. Adopting continuous practices and automation was seen as an opportunity to improve the situation. The interview findings indicate that while continuous integration has been widely adopted, deployment activities are still done manually in many cases. The identified barriers include organizational inertia, physical challenges in the operating environment, safety issues, and concerns about suspending production during deployment. A proof-of-concept design and implementation is presented, demonstrating a deployment solution for a ROS2-based application for use in simulation and autonomous mobile machinery. The solution aims to automate the deployment process of the distributed software architecture, following best practices and addressing relevant issues in the industry. After evaluation, the solution was deemed a promising design with automation significantly reducing deployment time and manual work.
  • Item
    Security Risks for Sidecar Containers in Kubernetes
    (2024-01-22) Halinen, Aarni; Bufalino, Jacopo; Martin Navarro, José Luis; Perustieteiden korkeakoulu; Di Francesco, Mario
    The widespread adoption of containers has been a driving force behind the shift from monolithic applications to microservices. The modular nature of microservice architecture has allowed the development of distributed system design patterns such as the sidecar pattern, in which peripheral tasks are separated from the application containers as their own distinct containers. The sidecar pattern is often used to implement observability, logging, and improving security with service meshes. This thesis examines the security risks associated with sidecar containers in Kubernetes container orchestration. It focuses on Kubernetes networking and Zero Trust security architecture to identify potential vulnerabilities that adversaries could exploit. The research reveals that Kubernetes only offers built-in security measures on the Pod-level, which can leave sidecar containers exposed to attacks. The thesis introduces two proof-of-concept solutions aimed at establishing network isolation between application and sidecar containers. The first solution utilizes IPTables to enforce isolation, while the second employs Multus to mimic shared network namespace between containers in distinct Pods. Despite the demonstrated solutions for network isolation, the complexities of implementation and the need for specialized infrastructure pose challenges that outweigh the benefits of such architectural modifications. Due to the challenges, the thesis explores the merits of sidecarless service mesh architectures such as Cilium and Istio Ambient Mesh, as an example of avoiding the risks with sidecars altogether. Ultimately, while the thesis presents viable proof-of-concept solutions for network isolation between application and sidecar containers, it advocates for a cautious approach, emphasizing the preference for established and well-tested components over custom-made implementations that need to be actively maintained.
  • Item
    The implications of technological neutrality for innovation policy design
    (2024-01-24) Carpen, Juho; Hakanen, Esko; Wolff, Joel; Perustieteiden korkeakoulu; Töytäri, Pekka
    In an ideal scenario, innovation policy leads to optimal societal outcomes by allowing the market forces to determine technology choices according to the principle of technological neutrality. However, criticism of technological neutrality has increased since 2000s as the market-oriented innovation policies have failed to address major societal challenges. This study contributes to this debate by examining the implications of the principle of technological neutrality for the effective design of transformative innovation policy mixes. The general research design of this study emphasises the approach of theory elaboration and abduction while including some traits of theory generation and induction. The empirical part of this study is conducted as an embedded single case study to analyse the role of technological neutrality in the Finnish innovation system. The final sample is a unique combination of 32 interviews, including a broad range of actors in the Finnish innovation system, and all 55 Finnish parliamentary discussions in 2014 – 2023 where technological neutrality was discussed. The data was analysed by following the principles of reflexive thematic analysis to identify and develop themes from the data. Finally, the empirical results were integrated into the theory-based models which were developed as part of the literature review. The main findings of this study are twofold. First, this study suggests that technological neutrality impacts innovation policy outcomes through three intermediate policy goals: (1) fairness, (2) technological goals, and (3) policy sustainability. The policy makers should pursue the desired goals by customising the policy design related to technological neutrality with different governance arrangements, policy goal boundary setting, and five different techniques to operationalise technological neutrality. Both technology neutral and specific policies can be justified depending on the desired goal and various contextual factors. Second, this study proposes a process for designing policy mixes from the perspective of technological neutrality by connecting the implications of technological neutrality into broader policy mix context. In conclusion, this study recommends taking a pragmatic approach to applying technological neutrality in policy design.
  • Item
    Self-Supervised Learning for Colloquial Finnish
    (2024-01-22) Sõrmus, Birgit; Getman, Yaroslav; Perustieteiden korkeakoulu; Kurimo, Mikko
    Self-supervised learning has shown great results for automatic speech recognition (ASR) systems as they utilize untranscribed speech data in their training, learning speech representations from raw audio data. In this work, the aim was to create a self-supervised speech model for colloquial Finnish and compare the results of pretraining from scratch and continued pretraining. To achieve this, three different models following the base Wav2Vec 2.0 architecture were pretrained. One model was pretrained from scratch while the other two were continuously pretrained. For the continued pretraining experiments, a monolingual Finnish model and a multilingual model were used. The experiments showed similar results for the model pretrained from scratch and the continuously pretrained monolingual Finnish model, with both yielding promising results. The model pretrained continuously from the monolingual Finnish model slightly outperformed the model pretrained from scratch, obtaining a word error rate of 28.1% and a character error rate of 7.5%.
  • Item
    Cortical activation in patients with alcohol use disorder during cue-reactivity task and C-tactile skin stimulation: An ERP study
    (2024-01-22) Puisto, Anna-Helena; Harkki, Juliana; Niemelä, Linda; Perustieteiden korkeakoulu; Renvall, Hanna
    Alcohol use disorder (AUD) is a highly prevalent mental disorder, with a cross-national lifetime prevalence of almost 10 %. Despite its widespread occurrence, a significant number of individuals struggling with AUD do not receive the help they need. The heterogeneity and complexity of AUD contribute to the challenge of developing an effective treatment for all. One potential solution for therapeutic intervention may involve activation of C-tactile (CT) afferents, as the CTs affect the addiction circuits in the brain. Event-related potentials (ERPs) are electrical brain responses that may provide promising biomarkers of the vulnerability to relapse in AUD patients. Earlier studies have shown that AUD patients express generally diminished amplitudes of the so-called P3 component. Furthermore, patients who express smaller P3 amplitudes elicited by alcohol-related than neutral stimuli are more likely to remain abstinent. This Thesis examined how CT activation modulates ERPs evoked by a cue-reactivity task in patients with AUD. The subjects were divided into an active group receiving CT-optimal skin stimulation and a control group receiving non-CT-optimal skin stimulation. During the cue-reactivity task, brain activity was measured with electroencephalography (EEG). After each neutral and alcohol-related image, subjects responded to questions related to the stimulus. The cue-reactivity values on the largest positive peak value between 300 ms and 600 ms (the P3 component) and on the mean ERP within the same time window between the groups were investigated. The statistical analyses conducted with Welch’s t-test at the individual sensor level identified a significant difference in the cue-reactivity values on the P3 component at the P3 electrode, and on the mean amplitude at the C4 electrode. Together, these findings suggest that CT-optimal activation may have a small modulating effect on ERPs in patients with AUD. However, to validate the results, further research involving more participants and exploring the cue-reactivity task between healthy and AUD patients is needed.
  • Item
    Exploring international students' needs and experiences with an online symptom checker in Finland
    (2024-01-22) Qi, Jiawei; Kujala, Sari; Perustieteiden korkeakoulu; Kujala, Sari
    Seeking guidance about health problems using digital health services has been an increasingly common phenomenon. Improved usability of online symptom checkers contributes to higher user acceptability and satisfaction. People from different countries have experiences and preferences varying with culture. However, concerns and needs of international students in Finland who are a young and highly educated group, but live in linguistically and culturally diverse contexts remain open for investigation. Thus, this study aimed to explore international students' needs and experiences in Finland with online symptom checkers. For this purpose, we conducted an exploratory case study about the Klinik Access symptom checker, and recruited six participants from several countries. In the usability testings, we adopted multiply data collection methods, including observation, questionnaires, and semi-structured interviews. The collected quantitative data was analysed using descriptive statistics, and qualitative data was analysed using thematic analysis and open coding method. As a result, we evaluated perceived usability using System Usability Scale (SUS), observed issues and perceptions on interface design. We summarized the facilitating and inhibiting factors affecting the use by international students. Based on the findings above, eight design suggestions were finally concluded. This will help to further optimize the design and development of services to make them more universally usable and more accessible to everyone.
  • Item
    Use of Plastics in Textile Industry - Exploring Stakeholder Perceptions, Innova-tion Challenges, and Pathways to Implementation
    (2024-01-24) Pukkala, Helena; Hakanen, Esko; Perustieteiden korkeakoulu; Töytäri, Pekka
    The textile industry, known for its significant environmental impact, is pressured for a shift towards sustainability. This transformation is driven by increasing regulation and consumer demand for more sustainable practices. The importance of sustainable fibre choices is also highlighted. However, there is no current concrete knowledge about different fibres' environmental, societal, and economic impacts. This knowledge gap is pivotal as the industry navigates towards a more sustainable future, necessitating a unified approach to fibre sustainability. This thesis examines the use of plastics and synthetic materials in the textile industry. The study involved conducting interviews with 22 individuals, encom-passing a wide range of industry professionals and textile consumers. These par-ticipants represent various segments of the textile supply chain and come from diverse consuming backgrounds, offering a comprehensive view of the industry's practices and challenges related to synthetic materials and their use. The findings of this thesis reveal that stakeholder perceptions and knowledge barriers influence material usage and innovation within the textile industry. The paper further implies that collaborative efforts among stakeholders are crucial for systemic changes promoting more sustainable practices. This research offers organisations insights and thus performs more informed decisions regarding material choices. However, the crucial need for a holistic approach, integrating regulatory, consumer, and sustainability perspectives to achieve improvements in the textile industry is underscored in the findings.
  • Item
    Network Security Policies for Containers in Cloud Applications
    (2024-01-22) Verma, Vishal; Peylo, Martin; Perustieteiden korkeakoulu; Aura, Tuomas
    This thesis delves into implementing network security policies within the application namespace of a Kubernetes cluster. As containerization and reliance on cloud environments continue to grow, robust network security becomes increasingly crucial. This research investigates the utilization of a specific port scanning tool in a Kubernetes cluster and utilizes a Container Network Interface (CNI) plugin to enforce network policies derived from the tool’s scanning results. The study offers an in-depth background on containerization, cloud applications, Kubernetes, and network security. Through a literature review, it highlights previous studies and approaches in the field, identifying existing challenges and gaps. The problem statement defines the necessity for an effective solution to enhance network security policies within containerized cloud environments. To address this, a test environment comprising a Kubernetes cluster and the chosen port scanning tool is established. The selected CNI plugin enforces network policies generated by the tool based on port scanning outcomes. The proposed solution undergoes evaluation and validation via experiments and analysis. This study contributes to the expanding knowledge base on network security in containerized cloud applications. Organizations can fortify their network security policies by employing the port scanning tool and CNI plugin, thereby reducing potential risks in their cloud-based environments. While applicable to various Kubernetes clusters, its primary focus remains the implementation and optimization of security measures within the MXIE, Nokia’s Kubernetes cluster tailored for Edge ecosystems. The insights derived from this research provide valuable guidance and practical implications for practitioners and researchers involved in cloud security and containerization domains.