25096 research outputs found
Sort by
Three-stage medical few-shot classification based on adaptive regularization with HMCE loss
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 medical research, the scarcity of labeled data and the high cost of expert annotation present a significant challenge for developing robust classification models, particularly in the context of rare diseases or specialized imaging modalities. To overcome this issue, we propose a three-stage few-shot learning framework that integrates meta-learning with pretraining and fine-tuning. First, during the pretraining stage, we pretrain the feature backbone on labeled external data using supervised loss to learn general feature representations. In the meta-training stage, we replace the fully connected layers of the pretrained model with task-specific fully connected layers and fix the feature extraction parameters. We then meta-train the fully connected layers on labeled simulated tasks using an adaptive learning rate and adaptive regularization with Hard-Mining loss, enabling rapid adaptation to new tasks. Finally, during the target task, we fine-tune the model on the target data, adjusting model parameters to align with the task’s feature distribution. We conducted experiments on challenging medical benchmarks BreakHis and ISIC2018 for few-shot classification tasks. Our method achieves superior performance on medical datasets, significantly outperforming related works. Additionally, ablation studies have also been conducted to validate the effectiveness of each module within the model
Relations with no one: a political-ethical pronominal analysis of the rise and rise of technological attachment figures in the form of dolls, robots, AI, Large Language Models and avatars
No full textUtilizing Force and Displacement in Unnatural Index Finger Movements for Authentication
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.The evolution of sensor technologies and real-time data processing has amplified the practicality of incorporating behavioral characteristics within security frameworks. Keystroke dynamics, in particular, has emerged as a prevalent behavioral biometric owing to the ubiquitous use of devices like mobile phones and computers, all reliant on password-based security systems. This study unveils an innovative authentication framework using leveraging deep learning algorithms, tapping into force and displacement data derived from the intricate abduction movements of the right index finger as a distinctive biometric trait. To ascertain its efficacy, we meticulously optimized this novel algorithm while benchmarking it against established deep learning models—Convolutional Neural Networks (CNNs), Gated Recurrent Unit (GRU), and Long Short-Term Memory (LSTM), and one-dimensional CNN (1D-CNN). The subsequent evaluation encompassed a comprehensive comparative analysis of their performancemetrics.Thefindings of this evaluation are compelling, demonstrating an average F1 score of 75.5% in validation data alongside an impressive average accuracy rate of 99.4%. These outcomes unequivocally highlight the precision and reliability inherent in utilizing force and displacement patterns as behavioral biometrics. Equally noteworthy is the system’s display of a remarkably low False Acceptance Rate (FAR) of 0.27%, positioning it as a promising contender for seamless integration within advanced security systems. In essence, this research not only showcases the potential of leveraging nuanced behavioral traits but also emphasizes thepracticalityandrobustness ofemploying force anddisplacement patterns as precise indicators in the realm of behavioral biometrics for enhanced system authentication and securit
Unveiling the poroelastic evolution of agar hydrogels through the drying process
No full textopen access articleAgarose-based hydrogels are widely used in food technology, biotechnology, and biomedical engineering owing to their biocompatibility, tunable microstructure, and well-defined gelation properties. Despite their extensive, the evolution of their mechanical properties during drying remains insufficiently understood. In this study, we investigate the time-dependent mechanical response of agarose hydrogels subjected to uniaxial compression during controlled ambient drying. Agarose gels of varying concentrations were prepared and characterized using compression testing, dimensional analysis, and microstructural observations.
The results reveal a non-monotonic evolution of the Young’s modulus during drying. In the early stages (within the first 24 h), a slight but reproducible decrease in stiffness is observed, which is attributed to mechanical instability of the semi-flexible polymer network induced by shrinkage. At longer drying times (24–72 h), continued water loss leads to pore collapse, network densification, and a pronounced increase in stiffness. These observations indicate the presence of two competing mechanisms governing the mechanical response: network buckling at short drying times and pore buckling at extended drying times.
By correlating macroscopic mechanical measurements with microstructural evolution, this work provides a mechanistic framework for understanding the drying-induced mechanical behavior of agarose hydrogels. These findings are relevant for the design and optimization of agarose-based materials in applications where controlled dehydration and mechanical stability are critical
Net Zero and Clean Energy Collaboration in regions without a combined authority: the case of Leicester, Leicestershire and Rutland
No full textA presentation highlighting challenges and opportunities in developing clean energy projects based upon local area energy planning analysis in the context of English regions without a combined authority structure
Mitigating cost estimation challenges in Bahrain's construction industry: a problem-solving approach
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.Purpose
In Bahrain's fast-growing yet under-researched construction industry, inaccurate pre-contract cost estimation has contributed to financial instability by increasing risk exposure and reducing project viability. Given the absence of localised guidance and the distinctive market dynamics of the region, this study aims to identify the key factors influencing cost estimation accuracy and to develop a practical framework to mitigate these issues.
Design/methodology/approach
A mixed-methods design was employed, consisting of a literature review, a survey of 49 construction professionals and a focus group discussion with five senior experts. Survey data were analysed using descriptive statistics, supported by a non-parametric significance test to assess robustness. The focus group discussion was transcribed and thematically analysed to validate and contextualise the quantitative findings. Insights from both phases informed the development of the proposed framework.
Findings
Estimation accuracy is primarily affected by the quality of client-provided information, the competency and experience of the estimation team and external market volatility. The study underscores the need for comprehensive project documentation and skilled professionals, and proposes a structured framework designed to enhance estimation reliability in Bahrain's construction sector.
Originality/value
This research addresses a significant gap in the literature by focusing on Bahrain, an under-examined yet strategically important construction market. The study contributes original empirical evidence on pre-contract estimation challenges and introduces a context-specific, expert-informed framework. The use of triangulated mixed methods enhances the credibility of the findings, offering practical value to industry stakeholders and advancing academic understanding of estimation practices in emerging markets
Optimising Harbour Construction Projects for Environmental Sustainability: A Hybrid Artificial Intelligence Approach
No full textopen access articleHarbour sedimentation represents a major challenge to the environmental sustainability and operational efficiency of coastal infrastructure, as frequent dredging activities increase maintenance costs, ecological disturbance, and carbon emissions. Conventional physical and numerical sediment transport models, while widely applied, are computationally intensive and often unsuitable for early-stage, sustainability-oriented design optimisation. To address these limitations, this study proposes a hybrid artificial intelligence-based optimisation framework integrating Artificial Neural Networks (ANNs), Genetic Algorithms (GAs), and Particle Swarm Optimisation (PSO) for sustainable breakwater and harbour layout design. Hydrodynamic simulations using the Coastal Modelling System (CMS) were conducted to generate a comprehensive dataset describing sediment transport behaviour under varying geometric and structural configurations. An ANN surrogate model was trained to capture nonlinear relationships between breakwater parameters and accumulated sedimentation volume, while GA-based global optimisation and PSO-based validation and local refinement were employed to identify optimal design solutions. Comparative assessment demonstrated consistent convergence of ANN–GA and ANN–PSO solutions within the same design region, with a maximum deviation of 8.46% between design variables and a sedimentation difference of 2.4%. The hybrid ANN–GA–PSO framework achieved the lowest predicted sedimentation volume, representing an improvement of approximately 2.3% relative to the ANN–GA baseline. The proposed framework supports Integrated Coastal Structures Management (ICSM) by enabling proactive, design-stage reduction in long-term sediment accumulation and dredging requirements, offering a scalable pathway toward sustainable and digital-twin-enabled harbour planning
Pilot of the DESNZ ‘Fit For Finance’ Tool
No full textIn May 2025, the Department for Energy Security and Net Zero (DESNZ) in the UK introduced a new online tool, named “Fit-For-Finance”, to aid local authorities in assessing their readiness for financing net zero projects. The tool supports self-assessment of readiness for implementing new financial schemes using private finance, blended finance, crowdfunding and other finance mechanisms to deliver infrastructural projects in the climate change space. The online tool comprises of 150 questions, covering 7 overarching themes, with 53 sub-themes defining a wide array of focused sectors for assessing financial readiness and innovation for local authorities. This report offers an illustrative account of the tool’s application and outputs, using Leicestershire County Council (LCC) as a case study local authority and drawing upon insights and data available through the Leicestershire Collaborate to Accelerate Net Zero (LCAN) project, funded by Innovate UK
Wastewater seasonal characterization and evaluation of reclamation potential of biological nutrient removal processes: a case study in Türkiye
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.This paper investigates the performance of biological nutrient removal (BNR) processes for wastewater reclamation potential. A case study of three conventionally designed BNR plants in Kocaeli, Türkiye evaluated the seasonal profile of their nutrient removal efficacy and potential reduction of their effective discharge into the receiving water bodies. The key parameters studied included - chemical oxygen demand (COD), biochemical oxygen demand (BOD), suspended solids (SS), total nitrogen (TN), total phosphorus (TP), and heavy metals. The analysis showed that effluent COD, BOD, SS, TN and TP concentrations of all three BNR plants were below 90, 20, 21, 9.2 and 1.2 mgL− 1 respectively, in compliance with the discharge limits set by the Turkish Water Pollution Control Regulations and the United Nations Food and Agricultural Organization for all the seasons. Further statistical analyses of the samples indicated that the data from all three BNRs followed a normal distribution, with p values generally greater than 0.05. Within the scope of analysis, no statistically significant differences were detected among the three BNRs. Beyond the current use of wastewater from the three plants for agricultural usage, further integration of dedicated wastewater recovery units is recommended alongside conventional BNR processes for enhanced wastewater reclamation for industrial and municipal applications
Artificial Intelligence in the Informal Economy: game changer for microentrepreneurs?
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.Purpose: This paper integrates insights from bricolage theory and dynamic capability framework to explore the potentialities and dangers of artificial intelligence (AI) in the informal sector, where microenterprises could harness its powers to transform their business models and scale, or risk falling further behind in the wake of AI-enabled disruption.
Methodology/approach: This paper takes a conceptual approach complemented with case illustrations. In the first part, it draws on bricolage and dynamic capabilities theories to introduce nine new propositions that explicate the dynamic, sometimes bidirectional, relationships, between AI, digital bricolage, dynamic capabilities and enterprise growth and competitiveness. In the second part, it highlights three illustrative cases of microenterprises to further elucidate these relationships.
Findings: This study proposes a novel framework integrating AI, digital bricolage, and dynamic capabilities (DC) to enhance the performance of informal microenterprises. It highlights the role digital bricolage as a mechanism for adapting existing resources to develop AI capabilities, and the complementary role of DC in deploying AI for growth, scaling and competitiveness. The study demonstrates AI's role in strengthening opportunity sensing, seizing, and transformative capacities that differentiates struggling enterprises from thriving ones, while also addressing critical limitations such as infrastructural inequities and fragmented skills.
Original contribution: This study proposes a novel framework integrating AI, digital bricolage, and dynamic capabilities (DC) to explicate the mechanisms and processes through which informal microenterprises achieve differential outcomes that propel some microenterprises to growth and scaling, on the one hand; while leaving others to fall further behind. To the best of our knowledge, this is the first paper that aims to unpack the double edged sword of AI as both a potential leveller and stratifier in the informal sector.
Practical implications: The study offers valuable practical implications for fostering inclusive digital transformation in informal microenterprises. It highlights the role of digital bricolage in enabling resource-constrained entrepreneurs to creatively adapt and deploy AI for value creation, operational efficiency, and agility. Policymakers and practitioners can leverage these insights to address barriers such as infrastructural inequities and skill gaps, fostering AI adoption. This approach supports sustainable competitiveness and market integration for marginalised enterprises
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!