27227 research outputs found
Sort by
DetDSHAP: Explainable Object Detection for Uncrewed and Autonomous Drones With Shapley Values
Get PDFAutomatic object detection onboard drones is essential for facilitating autonomous operations. The advent of deep learning techniques has significantly enhanced the efficacy of object detection and recognition systems. However, the implementation of deep networks in real‐world operational settings for autonomous decision‐making presents several challenges. A primary concern is the lack of transparency in deep learning algorithms, which renders their behaviour unreliable to both practitioners and the general public. Additionally, deep networks often require substantial computational resources, which may not be feasible for many compact portable platforms. This paper aims to address these challenges and promote the integration of deep object detectors in drone applications. We present a novel interpretative framework, DetDSHAP, designed to elucidate the predictions generated by the YOLOv5 detector. Furthermore, we propose utilising the contribution scores derived from our explanatory model as an innovative pruning technique for the YOLOv5 network, thereby achieving enhanced performance while minimising computational demands. Lastly, we provide performance evaluations of our approach demonstrating its efficiency across various datasets, including real data collected from drone‐mounted cameras and established public benchmark datasets
Conspiracy Theory as Metaphor: Promoting Cancer Misinformation through Algorithmic Influence and Metaphoric Manipulation on TikTok
Get PDFIllness has long been framed through metaphor. Whereas the media was once a primary vehicle to represent disease metaphors, new media technologies and short video platforms are increasingly used to represent health and illness through metaphor. This study (n = 200) employs an ethnographic approach to understand how metaphors are used to spread health misinformation online. It examines how cancer related health misinformation is circulated on the short form video app, TikTok. TikTok’s technological affordances encourage misinformation by providing accessible routes to monetisation and algorithmically steering users to content creators advertising unproven cancer cures. While TikTok’s recommender algorithm surfaces an abundance of health misinformation, purification and military metaphors play an important role in depicting cancer as a toxin that requires cleansing and detoxification by various unproven and harmful products and services. By emphasising the role of metaphor in spreading false and misleading information, this research highlights that misinformation is both a social and technical problem
Numerical Investigation of the End-Post Behaviour of Cellular Beams
Get PDFThe end-post, the solid web region between the end of the beam and the first web opening, is treated in a similar manner to the beam’s web-post in lack of definitive design guidance. However, its behaviour may vary depending on the adjacent connection type and the beam’s end configurations. This paper explores in detail the endpost behaviour of cellular beams, focusing on its failure modes and shear resistance. Building upon the experimental work by Tsavdaridis et al. (2024), three cellular beam setups were modelled and analysed using the FE software Abaqus. The FE modelling is discussed. Various end-post configurations were examined through a comprehensive parametric study. The FE analysis results, in agreement with the experimental results, yielded significantly higher failure loads than the design estimates given by Eurocode 3 Part 1-13. The stress distribution plots revealed the high stress concentrations at the vicinity of the first opening that initiated each failure mode. The connection type and the presence of notches or infill plates proved to dictate the stress distribution, ultimately controlling the failure mode
Simultaneous EEG-fMRI reveals a visual working memory encoding network related to theta oscillatory activity in healthy subjects
Get PDFWorking memory (WM) is crucially involved in many aspects of higher cognitive functions and goal-directed behavior. The encoding of sensory information necessitates the conversion of sensory stimuli into maintainable constructs. Oscillatory activity in the theta frequency range (4–8 Hz) of the human electroencephalogram (EEG) has been related to this. However, so far, no study has investigated the neurophysiological mechanisms and the brain network structure underlying the WM encoding process simultaneously. Thus, this study aimed to test whether theta oscillatory activity would be specifically related to the activity within a WM encoding brain network in healthy subjects by means of simultaneous recordings of EEG and functional magnetic resonance imaging (fMRI). Simultaneous recordings of EEG and fMRI were conducted in 32 healthy subjects during the performance of a visual working memory delayed matched to sample task. The fMRI analysis was informed by single-trial theta oscillatory responses to encoding stimuli. This analysis revealed a working memory encoding network mediated by theta oscillatory activity. The network included regions within the dorsolateral prefrontal cortex and parietal areas. Our results give reason to assume that the formation of a working memory network might take place during the encoding of information utilizing theta synchrony as a binding mechanism
Triply Periodic Minimal Surfaces Casing Treatment on an Axial Flow Fan: A Computational Aeroacoustic Study
Get PDFAxial flow fans are essential for cooling processes in various industrial settings, making noise reduction a vital factor. Initial studies indicate that adding a porous structure to the fan casing can effectively reduce noise caused by tip leakage vortices. Expanding on this idea, a porous casing based on Triply Periodic Minimal Surfaces (TPMS) was used to assess the fan's aeroacoustic performance. This study introduces the first application of TPMS casings for axial fan tip-leakage noise control. A thorough experimental and numerical analysis was carried out to investigate the aerodynamic properties and identify the mechanisms behind noise reduction. The findings show that the porous casing notably diminishes the strength and size of the tip leakage vortex while also decreasing turbulence intensity near the casing walls. Additionally, the analysis indicates that the degree of noise reduction depends on both the porosity percentage and unit cell size. A diamond TPMS structure with 55% porosity and a unit cell size of 6 mm provides an optimal configuration, enhancing viscous damping and sound energy absorption at low frequencies. This well-designed TPMS structure can effectively reduce both blade passing frequency and turbulence interaction noise, achieving an overall noise reduction of up to 10 dB
Comics & AI: Critical Prompts. A Multidisciplinary Conference on Comics, Technology and Creativity. Book of Abstracts & Bios
Get PDFComics & AI: Critical Prompts, A one-day multidisciplinary conference on the future of comics, technology, and creativity, took place on Thursday, 4 September 2025 at City St George’s, University of London. The file compiles in PDF form the conference schedule, the presentation abstracts and speakers bios that were also available on the page on the conference web site. https://comicsandai.org/schedule-abstracts-bios/. It includes the logos of the sponsors and a list of acknowledgements
Training and competency frameworks used in the preparation of healthcare professionals for head and neck cancer screening: A scoping review protocol
Get PDFObjectives
This scoping review aims to identify existing training and competency frameworks for healthcare professionals who assess or screen patients on the suspected Head and Neck Cancer (HNC) pathway or in extended practice models, and to explore how skills are developed, and competence assessed, prior to practice.
Introduction
Up to 97% of people referred with suspected HNC in the UK do not have cancer. This contributes to long wait lists and referral to treatment time in HNC services. A potential solution is for Speech and Language Therapists (SLTs) to assist such services by screening low-risk patients, specifically those presenting with hoarseness or oral-pharyngeal dysphagia. If SLTs are to screen low-risk patients safely and effectively, then specific training and competencies are required.
Inclusion criteria
International sources written in English between 2014–2025 that involve training and competencies for healthcare professionals who assess or screen patients for HNC.
Methods
The protocol presents the scoping review process that will be undertaken, which will be conducted as per the Joanna Briggs Institute Manual for Evidence Synthesis and reported in accordance with PRISMA-ScR guidance. Five databases and grey literature will be searched using a peer-reviewed search strategy. Professional organisations will be contacted for unpublished tools from clinical practice. Titles and abstracts will be screened using an a priori protocol. Eligible sources will be charted using an a priori framework and will undergo deductive content analysis. Results will be reported quantitatively and qualitatively using flow charts, tables and visual representations to outline the process and present the results
Towards Holistic Stakeholder Engagement: Integrating Participation, Inclusion, and Democracy
Get PDFResearch on the manifestations and hurdles of stakeholder engagement remains scant and fragmented. In this introduction to the Special Issue on stakeholder engagement, we conceptualize the concept through three related modes: participation, inclusion, and democracy. Each mode reflects different aspects of stakeholder engagement and can generate benefits for both organizations and their stakeholders. Yet, each mode also faces challenges. We propose an integrative framework that unifies these modes of stakeholder engagement, emphasizing areas of synergy and potential conflict. The article concludes by advancing directions for future business and society research on stakeholder engagement
Access Repository Dashboard
Do you manage City Research Online? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!