39449 research outputs found
Sort by
State death in buffer systems: Balancing and bandwagoning
Get PDFMaster's Thesis in Public AdministrationAORG351MASV-PUBA
The researcher as instrument - how our capacity for empathy supports qualitative analysis of transcripts
Get PDFIn this article we draw on literature from philosophy, history, and psychology to argue that empathy supports qualitative analysis of transcripts in several ways. We discuss examples of these processes both with data that we have co-created with our participants and data where we have not interacted with participants during data collection. What we suggest is not a new approach to analysis. Rather, we argue that the deliberate use of empathy bears potential to strengthen analysis across various analytical approaches. We explore five examples of how to access and harness our capacity for empathy as a resource in qualitative analysis: 1) Make time and room for prolonged engagement with data; 2) Use details and context actively when developing your understanding; 3) Practice decentering by actively seeking the perspective of the participants; 4) Attend to your own visceral experiences and body sensations; and 5) Utilize your capacity for imagination and creativity.publishedVersio
Approaching lateralized depression, a meta-analysis of the correlation between major depressive disorder and lateralized neural activity
Get PDFMasteroppgave i psykologiMAPSYK360MAPS-PSYKINTL-JUSINTL-SVINTL-MEDINTL-HFINTL-MNINTL-KMDINTL-PSY
Quantum Neural Network Architectural Interchangeability as a Noise Mitigation Technique
Get PDFQuantum Machine Learning (QML) is an exciting and relatively novel field that stands
at the intersection of quantum computing and classical machine learning. Quantum com-
puting enables computations that are infeasible on classical computing hardware, given
that quantum-specific methods are constructed to take advantage of the state space
provided by superposition and entanglement. However, the fragile nature of quantum
information and the prevalence of noise in Noisy Intermediate-scale Quantum (NISQ)
devices pose large challenges for the practical realization of QML models.
This thesis seeks to exploit Quantum Neural Network (QNN) architectural changes
as a noise mitigation technique in QNNs, as opposed to using qubit redundancy for quan-
tum error-correction. In particular, this thesis explores the effect of quantum noise on
gradients in QNNs by considering how the directional similarity, gradient magnitude,
and Euclidean distance evolve as a function of error rate compared to the idealized con-
ditions. Through quantum simulations, we exhaustively analyze different variants of the
angle-encoded RealAmplitudes ansatz using different combinations of rotational gates
and entangling techniques. We simulate QNNs with four qubits in noisy environments
with all permutations of bit flip, phase flip, and depolarization error under five discrete
noise error rates to quantify the gradient evolution through randomly sampling multiple
unique points in parameter space. With the scope limited to the Iris flower dataset, find-
ings show that phase flip noise exhibits notably lower directional similarity to the idealized
noise-free gradient than other noise permutations, whilst still having the highest gradient
magnitude across all the model architectures considered. Additionally, the performance
on the considered metrics is predominantly consistent across the noise models and QNN
variations. Results indicate that the noise impact on a four-qubit QNN evaluated on the
Iris dataset under different combinations of bit flip, phase flip, and depolarization noise
does not benefit greatly from changes to the network architecture as a noise-mitigation
technique. This work contributes to the broader understanding of noise-aware design in
QML and offers insight into alternative noise mitigation techniques.Masteroppgave i informatikkINF399MAMN-PROGMAMN-IN
Hospitality, Humanitarianism & Fragmented Governance An Ethnographic Study of Local Responses to Migration in Bihac, Bosnia & Herzegovina
Get PDFMasteroppgaveSANT350MASV-SAN
Korleis arbeider lærarar for å støtte elevar som står i fare for fråfall, og korleis vurderer dei den utvida læringsarenaen HUB i dette arbeidet?
Get PDFMasteroppgave i spesialpedagogikkSPED395MAPS-PEDMAPS-SPE
From TikTok to the Classroom A qualitative study of how high school students’ media use influences their understanding of the Israel-Palestine conflict
No full textFrom TikTok to the Classroom
A qualitative study of how high school students’ media use influences their understanding of the Israel-Palestine conflict
This master’s thesis examines how high school students’ media usage influences their understanding of the Israel-Palestine conflict and the didactic implications this has for teaching controversial topics in the Religion and Ethics (RE) curriculum. The primary objective is to map how students acquires knowledge about the conflict and identify the sources they use, with particular focus on the differences between traditional media and social media platforms, and what didactic implications this has for teaching controversial topics in the subject of religion and ethics.
To achieve this, a qualitative research design was employed, collecting data through focus group interviews at four high schools in Bergen. These interviews were structured to capture students’ reflections, knowledge sources, and experiences, providing insights into how media shapes their understanding. The study seeks to uncover various forms of knowledge, including emotional and opinion-based understanding, which extend beyond purely factual knowledge.
The findings may provide valuable insights for teachers who wish to address controversial topics in the classroom, contributing to more nuanced teaching that fosters critical thinking and democratic engagement among students. The study is therefore relevant for school leaders, educators, and others in the education sector.Religionsvitenskap mastergradsoppgaveRELV350MAHF-RELVMAHF-LÆH
Conceptualising research environments using biological niche concepts
Get PDFSeveral philosophers of science have taken inspiration from biological research on niches to conceptualise scientific practice. We systematise and extend three nichebased theories of scientific practice: conceptual ecology, cognitive niche construction, and scientific niche construction. We argue that research niches are a promising conceptual tool for understanding complex and dynamic research environments, which helps to investigate relevant forms of agency and material and social interdependencies, while also highlighting their historical and dynamic nature. To illustrate this, we develop a six-point framework for conceptualising research niches. Within this framework, research niches incorporate multiple and heterogenous material, social and conceptual factors (multi-dimensionality); research outputs arise, persist and differentiate through interactions between researchers and research niches (processes); researchers actively respond to and construct research niches (agency); research niches enable certain interactions and processes and not others (capability); and research niches are defined in relation to particular entities, such as individual researchers, disciplines, or concepts (relationality), and in relation to goals, such as understanding, solving problems, intervention, or the persistence of concepts or instruments (normativity).publishedVersio
Model Predictive Control for Hybrid Energy Systems
Get PDFThe global energy landscape has undergone increasing transformations in recent years, driven by the urgent need to adopt more sustainable and efficient energy sources. In Norway, offshore oil and gas installations have been the foundation of our society and a bedrock of our economy for the last five decades. Many of these now older systems still rely heavily on fossil-powered generators to provide stable and reliable energy for daily operations. They are significant contributors to the release of carbon on the Norwegian Continental Shelf (NCS). With new energy solutions constantly being explored and developed, previously remote systems, such as offshore installations, are altered to work as microgrids, incorporating renewable energy sources and laying the foundation for complex hybrid systems. These are connected to land and sea, from turbine installations to grid-connected megawatt systems. These systems are always searching for cost reductions and lower emissions. The optimization of their operation will help planners get insight into potential future developments and save costs. Many smaller renewable energy sources need to be controlled in parallel and yield the necessity for higher-order advanced control strategies and simulation tooling. As renewable energy increases within electrical systems, there are substantial opportunities for environmental and cost savings. Energy simulation tools have become important for assessing and optimizing energy use across various sectors, including building design, energy transportation, and industrial processes. These tools apply the modeling and analysis of distributed systems to enhance energy efficiency and reduce environmental emissions. We demonstrate how to build such energy systems in the modeling framework of Pyomo. Combined with Model Predictive Control, we show what an implementation for energy control can look like. We simulate a hybrid energy system that requires the continuous optimization processes offered by the control system. We show how such systems handle uncertainties regarding renewable energy production and price.Masteroppgave i informatikkINF399MAMN-PROGMAMN-IN
An Arrhenius reaction rate based burning model for simulation of dust explosions
Get PDFThe handling of combustible dust poses inherent risks, potentially leading to dust explosions whose strength and consequences hinge on various factors. Computational Fluid Dynamics (CFD) simulations of dust explosions can effectively incorporate these factors, contributing to the design of safer processes and equipment. The FLACS DustEx CFD code has demonstrated notable efficacy in simulating explosions in organic dusts such as maize starch, even in substantial structures like a 236-m3 silo. This code, formerly known as DESC, has been utilized by multiple industries for over a decade. The accurate functioning of the DustEx code relies on the combustion properties of the dust, influenced by factors like dust concentration and particle size. These crucial properties are derived from Pmax and (dP/dt)max values obtained through standardized experimental tests conducted in 20-L and 1-m3 environments.
The enhanced combustion modeling in DustEx comprises three key components. First, the burning rate is linked to an Arrhenius reaction rate based on the product temperature. Second, the impact of particle size distribution and dust concentration on the burning rate is considered by making it a function of the total surface area of all particles. Third, a substantial database at Gexcon is leveraged to formulate the burning rate model across a spectrum of dust types, particle sizes, and dust concentrations. These models are fine-tuned through optimization using least square fitting to align with experimental data.
In contrast to previous iterations of DustEx, which necessitated experimental tests for all combinations of dust concentrations and particle size distributions for each dust sample, the new models significantly reduce the need for such extensive testing. This innovation facilitates a more streamlined process, wherein experimental testing of dusts before CFD simulations can be minimized. The efficacy of these new models has been validated through simulations of large-scale explosion experiments, as well as experiments conducted in 20-L and 1 m3 spherical bombs. This comprehensive testing approach ensures the reliability and applicability of the improved combustion model across a range of scenarios and conditions.publishedVersio