25096 research outputs found
Sort by
Enhancing accessibility of laboratory learning environments for neurodivergent students
No full textAwareness and diagnosis of neurodivergent conditions among Higher Education (HE) students is increasing and consequently more are seeking support and reasonable adjustments. We will describe a cross-institutional collaborative project that is investigating accessibility of laboratory taught settings for neurodivergent students, as many current initiatives to enhance accessibility do not translate well into these learning environments. By conducting focus groups and case study interviews with self-identified neurodivergent students engaged in a Science, Technology, Engineering or Mathematics (STEM) course at either the University of Nottingham or De Montfort University, we aim to gain a deeper understanding of lived laboratory experiences of these students across multiple Schools/Faculties/Institutions.
We will share common challenges faced by neurodivergent students in laboratory learning environments and the insight gained regarding appropriate and reasonable adjustments that can enhance accessibility/inclusivity, experiences, and ultimately academic outcomes. We will also cover a key output of the project - to generate a set of guidelines for HE educators, that aspires to become a valuable resource raising awareness of the issues and ideas to effectively address them. We will be seeking feedback and implementation at additional institutions in order to maximise the impact and utility of the guide across the sector
Exploring skilled immigrants' access to decent work and the effectiveness of the role of third-sector organisations in the UK
No full textThis research explores skilled immigrants' access to decent work and the effectiveness of the role of third-sector organisations in the UK. It uses qualitative data from UK's TSO providing skills and employment services and skilled immigrants. The findings are expected to enhance understanding of decent work experiences of skilled immigrants in the UK and the contributions of TSOs
A multivariable grey prediction model with different accumulation operators and its applications
No full textThe file attached to this record is the author's final peer reviewed version. The Publisher's final version can be found by following the DOI link.Existing multivariable grey prediction models employ a uniform accumulation operator to preprocess data across variables, disregarding the dynamic relationships between each variable’s data characteristics and the functional structures of different grey accumulation operators. This study proposes a novel model that employs adaptive operator variability and dynamically adjustable orders, with accumulation operators tailored for each variable through combinatorial optimization. The model’s effectiveness and practicality are validated through two case studies and applied to predict energy consumption in Chongqing. The experimental findings indicate superior performance compared with other models. This study advances the methodological framework for multivariable grey prediction models
The impact of announcing a payment assistance program on UK household water consumption: A regression discontinuity analysis
Get PDFopen access articleInformation campaigns aimed at social norms are a mainstay tool for utilities. The announcement effect of such policies is usually ignored in favour of ex-post examinations of campaigns' outcomes. This paper examines the effects of announcing a ‘help-to-pay’ program from a United Kingdom (UK) water utility company for households facing a uniform price for water. This study uses data for around 6000 Yorkshire Water households between 2018 and 2020, adopting cross-sectional Regression Discontinuity (RD) and regression approaches. Leveraging household characteristics, social characteristics and weather-related information such as rainfall and temperature at the regional level, this study explains household water consumption behaviour on and around the announced £350 per year water bill threshold. Results indicate a discontinuity in annual household water consumption around the threshold, and announcing the eligibility criteria of the help-to-pay program results in higher water consumption for households meeting some criteria (11.9k litres/household/year). Through falsification tests, evidence exists of household strategic behaviour around the threshold. With water bills in the UK scheduled to increase by up to 40 % in the coming years, this research provides valuable insights into the effect that announcing bill support schemes can have and insights on potential strategic behaviour on behalf of households attempting to offset costs
Generating global high-resolution current and future weather timeseries, and the implications for India
Get PDFopen access articleBuildings are responsible for 39% of world carbon emissions, mostly from heating and cooling driven by the weather. Hence, successfully designing buildings for climate mitigation and adaptation is fundamentally dependent on good quality current and future weather timeseries. Unfortunately, in most of the world and especially in the Global South, where the impacts of climate change are predicted to be greatest, data with sufficient geographic or temporal resolution do not exist. Here we demonstrate a new globally-relevant method to produce high spatial-resolution, carefully calibrated, mutually consistent, hourly, current and future typical weather years using low-cost and high-quality synthetic data. The approach is then applied to India, which currently accounts for 18% of world population and is expected to add around 14% of all new buildings in the world by 2050. This results in an order of magnitude improvement in spatial resolution (~400 → 4,790 locations) for current (1981-2010) climate, with future (2060-2089) climate represented for the first time. Systematic comparison of several methods suggested multivariate kriging as ideal to scale spatio temporally sparse calibration data to all 4,790 locations. By moving a calibrated computer model of a typical home over all 4,790 locations, we find a mean increase of 3 K in indoor summer temperatures and a 51% increase in cooling demand by 2060 compared to 2010, and the potential for severe heat stress. Finally, we make these files free-to-use, thus creating the first such large-scale public repository for anywhere in the Global South, with potential use in many other fields such as infrastructure and crop resilience
The cost of “everything”: minoritized social entrepreneurs’ response and adaptation during the cost-of-living crisis
Get PDFThe file attached to this record is the author's final peer reviewed version. The Publisher's final version can be found by following the DOI link.Purpose
The cost-of-living crisis negatively impacted many parts of the United Kingdom, exposing the heightened economic failures faced by impoverished people whom social enterprises (SEs) support. Yet, the resilience of SEs, especially those led by minoritized individuals hardest hit by the crisis, is underexplored. This paper examines the response and adaptation of Black- and Asian-led SEs during the cost-of-living crisis. It draws from Duchek’s (2020) organizational resilience conceptualization to offer valuable insights into crisis response strategies and their implications.
Design/methodology/approach
A qualitative design is adopted for semi-structured interviews with SE founders and directors. The interview data were analyzed using abductive analysis.
Findings
Proactive, reactive and defensive strategies are the primary response approaches. Most proactive organizations are Asian-led, while reactive organizations are predominantly Black-led. There is an equal organization for the defensive strategy; however, various capabilities were used for each response approach. The response approaches and capabilities are determined after an introspection of the business models and performance. Thus, SE resilience combines capabilities specific to each crisis response approach.
Practical implications
Black- and Asian-led SEs showed no significant differences in their responses to the cost-of-living crisis. However, adopters of reactive and defensive strategies must establish a learning process to enhance preparedness for future crises and foster resilient systems.
Originality/value
The research constructs an SE response typology framework to expound on the behaviors of three organizational response categories: cost-driven innovators, market expansion defenders and innovation pioneers
Image denoising by attention U-Net based network module for automated enhancement of low light images
No full textEnhancement of low-light images is a challenging process since it considers brightness recovery, noise, and distortion. The images captured in low light conditions degrade the image quality and increase the complexity of future computer vision processes. To solve this problem, this work presents an improved deep learning-based image enhancement model for low-light images. This process goes through three main phases: image decomposition, noise reduction and image enhancement. Initially, the low-quality images from the datasets are decomposed using residual layers in the residual network, which obtain illumination maps by decomposing low-light images. After decomposing the images, the denoising process is carried out by using the attention U-Net model, which is used to suppress the noise and achieve better-quality images. Finally, the improvement network is employed to boost the images’ contrast and brightness. This enhancement network includes a convolution and deconvolution layer that handles feature map information from low-light images. Finally, May Flame algorithm optimization is used to reduce the loss function and improve the quality of the images. The performance metrics such as peak signal noise ratio (PSNR), similarity index measure (SSIM), Natural image quality evaluator (NIQE), Lightness Order Error (LOE), Optimal Scale Selection, OSS-PSNR, Universal Quality Index (UQI) will be computed and compared with the recent techniques. The proposed model can obtain PSNR values of 21.06, SSIM range of 0.9456, UQI range of 0.9874, LOE values of 241.86 and NIQE of 16.16. Thus, the proposed model is more efficient than other existing models
A CatBoost and ExtraTrees-Based Softvoting Ensemble Approach for Non-Invasive Diabetes Detection Using Hair LIBS Spectral Data
No full textThe file attached to this record is the author's final peer reviewed version. The Publisher's final version can be found by following the DOI link.Type II diabetes, a chronic metabolic disorder, alters the chemical composition and structural integrity of biological tissues, leading to widespread health challenges. Human biomarkers, such as hair, offer valuable insights into metabolic variations, making them useful for disease classification. Laser-Induced Breakdown Spectroscopy (LIBS), a rapid and non-destructive technique, presents a powerful tool for analyzing hair samples to distinguish between diabetic and healthy individuals based on spectral data combined with machine learning. This study integrates LIBS with advanced tree-based machine learning to advance classification accuracy. Specifically, we utilize the predictive strengths of CatBoost and ExtraTrees classifiers and combine them using a soft-voting ensemble approach. Hair samples from 120 volunteers, 50 healthy controls, and 70 diabetics were analyzed, yielding high-resolution emission spectra. Principal Component Analysis (PCA) was employed to extract key spectral features, facilitating efficient data interpretation. The integrated softvoting ensemble learning model achieved an outstanding classification accuracy of 97.5 %, significantly exceeding the performance of the standalone CatBoost (94.5 %) and ExtraTrees (92.8 %). To evaluate the real-world applicability of the model, it was further tested on an external dataset comprising 80 independently collected samples (45 diabetics and 35 healthy controls), achieving 96.2 % classification accuracy with consistent sensitivity and specificity. These results highlight the transformative potential of ensemble-based machine learning in LIBS-driven disease diagnostics, paving the way for more precise, non-invasive, and scalable diabetes screening methods
Experiment and simulation to investigate fatigue life of R260 of LRT
No full textThe file attached to this record is the author's final peer reviewed version. The Publisher's final version can be found by following the DOI link.In this study, an experimental and fatigue simulation is designed to investigate the fatigue life of R260 rail for light rail transit. Specifically have two parts in experiment, begin in tensile test to examine the value of UTS, yield strength and break elongation for R260 material behaviour using UTS machine below 250kN through three specimens. In second part, the type of Servo-Pulser fatigue machine was performed of less 100 kN with using frequency of 10Hz based on 20% to 80% separately as below value on UTS of 1145 MPa to identify fatigue life under stress ratio of R = −1 to generate S-N curve. The results of the experiment show that the endurance limit occurs at 20% and 30% stresses loading below 372 MPa, the regime of HCF fall into 40%untill 70% ranges at 103-105 cycle to failure and LCF occur of 80% ranges at 102 cycle to failure. In fatigue simulation, the results of fatigue limit is below 372 MPa of 20% and 30% and the fatigue failure fall into 103-105 cycles period in stress ranges in between 372 MPa to 763 MPa. Both results gives a good agreement correlation to prove feasibility and reliability in terms of creation R260 databases
Data sources:
Access Repository Dashboard
Do you manage De Montfort University Open Research Archive? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!