48651 research outputs found
Sort by
Active Deep Clustering: Exploratory Analysis to Assist in Decision-Making on Incremental Label Morphing Datasets
No full textMany supervised-based training schemes rely on the need to have a single associated label for each data sample within a set. Where the goal is to learn different levels of granularity of the data in an implicit form, in the context of neural networks, this is accomplished with different layers to extract features at distinct levels. In this work, we explore more explicit labelling structures where each sample has multiple labels forming relationships at both an abstract and fine-grained level, producing a tree for each associated data sample. This novel type of training scheme utilises a refinement strategy based on deep clustering approaches to detect the colliding and splitting of clusters where each is assigned a label. Experts can then be queried to determine if those colliding clusters should belong to a single label, or alternatively, if they are splitting, should we create new labels, forming an active component of our training scheme. Colliding clusters form a parent label while splitting clusters form sibling labels within the tree structure. By utilizing a tree data structure to represent labels at different levels of granularity, we can invoke explicitly defined relationships and dependencies to form a more structured and interpretable representation of data. Instead of treating data as flat, homogenous sets, we allow for the exploitation of hierarchical relationships and leverage inherent structure to improve data efficiency. We present a case study of the approach applied within the steel manufacturing domain, where quality control remains an active challenge due to morphing labels as products move down the production line
Outdoor Stability of 518 cm2 Active Area Screen-Printed Mesoscopic Carbon-Based Perovskite Solar Modules Over 12 Months
No full textMesoscopic carbon-based perovskite solar cells (C-PSCs) composed of screen-printed TiO2, ZrO2, and carbon layers offer a pathway to stable, scalable, low-cost photovoltaics via commercially mature fabrication methods. While their potential lifespan has been demonstrated under standardized conditions, few studies examine the behavior of large-area modules exposed to real-world environments. Here, 12 months of outdoor weathering data are presented for 518 cm2 active area MAPbI3 modules with over 80% geometric fill factor, fabricated using low-cost mechanical scribing. Modules exhibited power conversion efficiencies (PCEs) up to 9.4% under 1 sun, with PCE increasing at lower light intensities. Following outdoor continuous intermittent power point tracking for over 12 months, an encapsulated module retained 68% of its initial PCE. Performance remained stable during cooler months, only falling when temperatures rose during summer months. Similar temperature-dependent trends are observed in repeated trials. Weathering trials identified key degradation pathways linked to fabrication—namely, non-uniform heating during perovskite annealing, encapsulation, and infiltration-related failures. Controlling heat exposure and conformity during module manufacture and operation is therefore critical to extending lifetime. These results highlight the importance of real-condition assessments in optimizing the scale-up of novel perovskite technologies, providing key insights into the steps required to achieve commercially viable lifetimes
Integrating Rule-Based eGFR Labels with Expert GP Annotations: A Multi-method Framework for CKD Classification
Full text linkPredicting shoreline changes using deep learning techniques with Bayesian optimisation
Full text linkAccurate prediction of shoreline change is vital for effective coastal planning and management, especially under increasing climate variabilities. This study explores the applicability of deep learning (DL) techniques, particularly Long Short-Term Memory (LSTM) and Convolutional Neural Network-LSTM (CNN-LSTM) models, for shoreline forecasting at monthly to inter-annual timescales, under two modelling approaches—direct input (DI) and autoregressive (AR). All models demonstrated the ability to reproduce temporal shoreline variability, while the autoregressive DL models were performing better.Further, a noise impact assessment revealed that seasonal decomposition and noise filtering significantly enhanced the model performance. In particular, the models using 52-week data decomposition and residual noise reduction improved the model performance. The reduction of data noises also resulted in narrower ensemble prediction envelopes, indicating that ensemble candidate models behave with low diversity. The temporal data resolution analysis showed that lower data resolutions reduce the predictive performance of the model and at least fortnightly data are required to satisfactorily capture the trend of variability of the shoreline position at this beach.The use of ensemble predictions, derived from a selected subset of model trials based on their collective performance, proved beneficial by capturing diverse temporal behaviours, thereby offering a quasi-probabilistic forecast with minimal computational cost. Overall, the study underscores the potential of DL models, particularly with autoregressive architectures, for reliable and transferable shoreline change prediction. It also emphasizes the importance of data quality, resolution, and preprocessing in improving model robustness, laying the groundwork for future research into use of DL in multi-scale shoreline predictions
Gold Nanoparticle Melting: Effects of Size, Support Interaction, and Orientation
No full textAn understanding of nanoparticle (NP) melting is essential for both fundamental nanoscience and the design of high-temperature catalytic systems. We investigate the melting behavior of truncated octahedral gold NPs, ranging in size from 2 to 4 nm, supported on their edges, (100) or (111) facets, using molecular dynamics simulations, with a machine-learning force field trained on density functional theory data. We systematically examine the effects of NP size, support interactions, and orientational dependence by applying spring-like constraints to specific facets or edges. Our results show that NP melting follows the liquid nucleation and growth model, with surface disorder preceding rapid melting at a critical temperature. Constraining the atoms to simulate contact with a support consistently raises the melting temperature, with stronger effects for smaller clusters, and for (100) facets compared with (111) facets, that is, there is an orientational effect. Importantly, the extent of the offset in melting temperature is quite independent of the interaction strength, implying that all support interactions can significantly stabilize small NPs. These findings provide a framework for more accurate predictions of nanoscale melting in practical catalytic environments
Femilial Cartography: The Hybrid Aesthetic as Decolonial Practice
No full textThis thesis includes a creative work, Femilial Cartography, and a critical exegesis, The Hybrid Aesthetic as Decolonial Practice, both of which explore how the hybrid aesthetic functions as a decolonial methodology, challenging dominant epistemologies and offering alternative ways of narrating self, history, and community.Femilial Cartography, a formally innovative work, begins as an intimate exploration of a mother-daughter relationship and unfolds into a broader meditation on memory, identity, and inheritance. Through a fragmented, non-linear structure, the narrative traverses time, borders, family, history, to unearth stories buried by colonialism, migration, and gendered silence.Disrupting singular historical accounts, it asks how the legacies of empire shape national identity, cultural expectation, and personal memory.Assembling and interpreting a family archive, I uncover subjugated histories and reconstruct personal and collective memory, tracing how ordinary lives are entangled with colonialism, Partition, migration, and inherited identity. This engagement becomes a way of challenging dominant epistemologies rooted in colonial logics. Through opacity, affect, and the fragment, my creative methodology resists these frameworks, reimagining the archive as a living, unstable, generative space.Femilial Cartography also considers how memory is transmitted, fractured, and reshaped across generations, revealing the entanglement of family narratives with broader political forces. The work links the enduring influence of Indian colonisation to contemporary Pakistani norms around nationhood, motherhood, and the gendered expectation that women act as custodians of cultural identity.The critical exegesis theorises the strategies of the creative text, articulating a hybrid aesthetic as decolonial practice. Drawing on Édouard Glissant’s concept of opacity, it defends fragmentation, non-linearity as aesthetic refusals of dominant epistemologies and a disruption of narratives that demand clarity and categorisation. It introduces the new critical termfemilial to describe the nuanced mother-daughter dynamic. Unlike familial, which generalises kinship ties, femilial emphasises the specifically gendered and affective dimensions of the relationship
Acquisition of competence: An analysis of clinical teaching and learning of midwifery at Kamuzu College of Nursing
Full text linkIntroduction: Malawi has one of the highest maternal and neonatal mortality rates globally. In response a competence-based education (CBE) approach in midwifery education was introduced at Kamuzu College of Nursing (KCN). KCN adopted the International Confederation of Midwives’ seven essential competencies for basic midwifery practice to produce professional midwives. However, there are reports that the performance of the graduates is below standard.Purpose: To explore the clinical teaching and learning practices utilized by midwifery lecturers and students at KCN in preparation of students for effective midwifery practice.Design: A sequential qualitative study was conducted at KCN in Malawi. Data were collected from multiple sources for triangulation. Purposive sampling was used to select a sample of six senior midwives and six educators in first phase. In the second phase, 26 student midwives, and five graduates from KCN, and for comparison, four graduate midwives from another local institution, Mzuzu University (Mzuni). Face to face semi structured interviews were conducted in the first phase. In the second phase focus group discussions were conducted to collect data from the students and graduates from KCN, and face to face interviews were used for Mzuni graduates to obtain participants’ accounts of the phenomenon. Timetables, the curriculum and students’clinical assessment forms were checked to verify data from respondents. Using NVivo software, a thematic analysis approach was used for data analysis.Findings: Findings reveal that learning is compromised. Although the curriculum document indicates that the program is competence based, teaching and learning methods, and assessment of students’ clinical learning are inconsistent with CBE,including learning theories such as cognitive load, situated learning, psychomotor skill learning and experiential learning theory. There is paucity of resources, poor infrastructure, and poor personal relationships.Conclusion: The study portrays a negative impact of introducing change using top management as a change agent
Freshwater producing technologies for active buildings in developing countries
Full text linkThe aims of this work were to assess freshwater producing technologies to look for potential improvements and to investigate a potential use for the water produced. To do this an assessment was undertaken of the potential for improving solar thermal desalination by coupling a solar desalination system with a solar thermal collector. The purpose of the solar thermal collector was to pre-heat the inlet air entering the desalination system to improve water productivity. To maximise air temperature uplift, low Solar Reflectance Index (SRI) steel panels, high transmissivity glass, and small air gaps were optimal. Low air flow rates produced higher air temperatures but reduced the useful gain of the collector. Higher air flow rates are also beneficial for improving evaporation rates, and thus maximising temperature at higher air flow rates would be optimal for desalination. At these higher air flow rates, the transpired solar collectors showed the highest air temperature uplifts. Simulating flowing the heated air from a solar collector through the desalination device to improve the evaporation was found to be inefficient and the desalination device struggled to produce water condensate without the water being heated by an external source. When controlling the water temperature, it was determined that the water productivity of the desalination device was 1.28 L/m2/day at 50°C water temperature. An alternative freshwater production method of moisture harvesting was also investigated. The lithium chloride hexagon was optimal at the highest relative humidities with 175% of its mass absorbed at 80% relative humidity. An aeroponics building that may utilise the freshwater produced was modelled for the electrical energy production to meet a set electrical load, using an off-grid system with solar PV-T. The cooling loads of the building were found to be significant, but these have potential to be met by PV-T and thermal storage
NEOM - Lining Up a New Smart City
Full text linkNEOM, a US$500 billion megaproject under development in Saudi Arabia and a flagship of the Kingdom’s Vision 2030, proposes a new model of urbanism powered entirely by renewable energy, governed through AI-driven systems, and anchored by The Line, a 170-kilometre linear city designed to eliminate car dependency, reduce carbon emissions, and preserve 95% of its surrounding natural environment. This study examines how the planning of the NEOM urban area draws upon lessons from other established smart cities worldwide, with an emphasis on NEOM’s planning motivations, environmental strategies, and its distinctive linear spatial logic. Boyd Cohen's Smart City Wheel is used as a diagnostic framework to assess NEOM’s alignment with established smart city dimensions, while the lens of smart and cognitive urbanism offers insight into the project’s symbolic, technological, and strategic dimensions. A qualitative interpretive case study methodology guided the research during NEOM’s construction, gaining empirical insights into NEOM’s formative planning and early construction stages. Primary data was collected through 39 semi-structured interviews with high-level internal stakeholders, including NEOM planners, consultants, and experts, and external participants such as urban scholars, sustainability professionals, and policy advisors. These insights are supported by documentary analysis and in-depth engagement with planning reports and media sources. The findings reveal that NEOM selectively integrates global best practices while extending them through novel frameworks, such as AI urbanism, Zero Gravity Urbanism, and cognitive dimensions. While the project introduces ambitious innovations in renewable energy systems, vertical spatial design, and AI-enhanced service delivery, it remains in an early experimental phase. At this stage, NEOM cannot yet be considered a fully replicable model, but it embodies a potential future paradigm marked by conceptual boldness and global significance. The study concludes by identifying key challenges for translating NEOM’s ambitious vision into a viable and replicable model, particularly in terms of inclusive governance, environmental resilience, technological feasibility, and the uniqueness of NEOM’s political and geographical context.Finally, recommendations for future research are suggested as NEOM continues to be built, specifically the need for longitudinal monitoring and comparative studies to evaluate and contextualise its evolving progress and replicability
Data sources:
Access Repository Dashboard
Do you manage Swansea University Cronfa? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!