26207 research outputs found
Sort by
Embedding-based detection of indirect prompt injection attacks in large language models using semantic context analysis
No full textLarge Language Models (LLMs) are vulnerable to Indirect Prompt Injection Attacks (IPIAs), where malicious instructions are embedded within external content rather than direct user input. This study presents an embedding-based detection approach that analyses the semantic relationship between user intent and external content, enabling the early identification of IPIAs that conventional defences overlook. We also provide a dataset of 70,000 samples, constructed using 35,000 malicious instances from the Benchmark for Indirect Prompt Injection Attacks (BIPIA) and 35,000 benign instances generated using ChatGPT-4o-mini. Furthermore, we performed a comparative analysis of three embedding models, namely OpenAI text-embedding-3-small, GTE-large, and MiniLM-L6-v2, evaluated in combination with XGBoost, LightGBM, and Random Forest classifiers. The best-performing configuration using OpenAI embeddings with XGBoost achieved an accuracy of 97.7% and an F1-score of 0.977, matching or exceeding the performance of existing IPIA detection methods while offering practical deployment advantages. Unlike prevention-focused approaches that require modifications to the underlying LLM architecture, the proposed method operates as a model-agnostic external detection layer with an average inference time of 0.001 ms per sample. This detection-based approach complements existing prevention mechanisms by providing a lightweight, scalable solution that can be integrated into LLM pipelines without requiring architectural changes
Robust finite-time control of multi-link manipulators: A data-driven model-free approach
No full textIn recognising both the emerging industrial applications of multi-link robotic manipulators and the inherent challenges of modelling and controlling their highly complex nonlinear dynamics, this work proposes a completely model-free terminal sliding mode control (MFTSMC) design approach to reduce the sensitivity and complexity often associated with model-based routines. Consequently, the proposed design achieves strong robustness, simplicity, and good operation tuning by eliminating the need for system modelling and enabling direct operator–machine interaction. Simulink simulations on a 3-link case subjected to different disturbance conditions (free, low-frequency, high-frequency, and mixed) show rapid dynamic convergence, good tracking precision, and strong disturbance rejection. The system reaches the sliding surface within 0.07 s, maintains steady-state errors around 10−2, and achieves a smooth torque response with low energy costs. The benchmark results confirm the finite-time convergence and demonstrate that the proposed framework is practical and scalable for multi-DOF systems and has potential for underactuated manipulators. It should be noted that a generalised dynamic model for a planar n-link manipulator is presented in the study for (1) the ground truth of the manipulator in simulation (not for the MFTSMC design), (2) the model-based controller designs in comparison to the MFTSMC, and (3) understanding the dynamic characteristics
Economic policy uncertainty and AI-driven stock spillovers: Implications for portfolio diversification
No full textThis study examines how economic policy uncertainty (EPU) shapes return spillovers between major AI-driven stocks. Using daily data for seven leading AI-driven firms, this study estimates a time-varying connectedness via the TVP-VAR framework and links it to macroeconomic uncertainty using a GARCH-MIDAS-EPU model. The pre- and post-COVID analyses capture structural shifts in spillover dynamics. The results show Microsoft and Google are primary transmitters of shocks with Nvidia acting as a critical intermediary, while Tesla and IBM act as net receivers. EPU amplifies short-term connectedness but dampens long-term spillovers, reflecting risk-averse behavior and delayed investment under uncertainty. In post-pandemic periods, long-term volatility persistence strengthens and connectedness becomes more sensitive to policy shocks. Further, the portfolio testing reveals that strategies minimizing correlation and connectedness outperform traditional variance-based approaches during periods of heightened uncertainty. These findings highlight AI-driven equities as a critical channel of macro-policy risk transmission and provide actionable insights for investors, portfolio managers, and policymakers in innovation-driven financial markets
Seeking digital maternity healthcare during the pandemic health system shock: A systematic review of women's experiences in low- and middle-income countries
No full textBackground: The pandemic created global disruption acting as a health system shock not seen before in living memory. As a consequence, there were significant implications for healthcare delivery in low- and middle-income countries. Challenges such as lockdown restrictions created substantial modifications to the delivery of maternity care. This review aims to explore the experiences of maternity care by women, specifically in low- and middle-income countries, during the pandemic global health system shock.Methods: A systematic search was conducted for qualitative literature published about maternity healthcare experiences during the pandemic. Studies which provided qualitative data on women's experiences of digital healthcare, and other maternity care reconfigurations in low- and middle-income countries were included. The studies underwent quality assessment using twelve criteria adapted from the quality appraisal tool developed by the Evidence for Policy & Practice Information (EPPI) Centre. Thematic synthesis was employed.Results: Of the 21,860 records identified, 30 met the inclusion criteria for this review. Across the 4 key predetermined areas of study: (1) Care seeking and experience; (2) Digital health; (3) Vaccination; and (4) Ethical future of maternity services; 10 concepts were reported upon, namely: (1.1) Emotional challenges and uncertainty, (1.2) Disruption of services, (1.3) Stigma and discrimination, and (1.4) Changing support systems; (2.1) Safety and reassurance, (2.2) Locus of responsibility; (3.1) Vaccine understanding and acceptance; and (4.1) Improvements for maternity care delivery, (4.2) Implementation of virtual care, (4.3) Education and empowerment.Conclusion: Our findings suggest emotional challenges, isolation, and limited access to maternity services were prominent among pregnant individuals in low- and middle-income countries. This synthesis provides insights into how pandemic associated adaptations, which have been retained beyond, such as digital health solutions were experienced by women within constrained health systems, revealing both opportunities and persistent gaps in digital health access and equity. Although a review of low- and middle-income countries—there is learning to be taken from these settings which could easily be applied not only across low- and middle-income countries, but also in high-income settings, in the form of reverse (or “trickle-up”) innovation to improve maternity care as we recover and re-build from the pandemic and offer more resilient ways of providing maternity care through future health system shocks
Antimicrobial effects of combined piezoelectric cold plasma, organic acids, and nanoemulsions against Salmonella Typhimurium and Listeria monocytogenes on pork
No full textThis study evaluated the antimicrobial efficacy of cold atmospheric plasma (CAP) alone and in combination with nanoemulsions or organic acids against pathogens on polycarbonate membranes and pork meat. CAP was generated via piezoelectric direct discharge technology with ambient air as the working gas. On polycarbonate membranes, CAP treatment alone for 15, 30, and 45 s reduced Salmonella Typhimurium by 0.9, 1.4, and 2.4 log CFU/cm2 and Listeria monocytogenes by 0.7, 1.7, and 2.3 log CFU/cm2, respectively, showing a time-dependent antimicrobial effect. When CAP was applied before lactic or acetic acid (at minimum inhibitory concentrations (MICs)/minimum bactericidal concentrations (MBCs)) on polycarbonate membranes, the combined treatments achieved significantly greater reductions (~ 3.6 log CFU/cm2) than when acids were applied before CAP (~ 2.3 log CFU/cm2), highlighting the importance of application sequence. Overall, CAP treatments on polycarbonate membranes showed additive effects when CAP (applied for 15, 30, or 45 s) was combined with antimicrobials (at MIC/MBC). On pork, CAP treatment for 9 min combined with organic acids or nanoemulsions at 10× MIC produced significant additive effects, enhancing pathogen inactivation (by ~ 1.5 log CFU/g) compared with CAP alone or antimicrobials alone under the same conditions. These findings support the application of CAP–antimicrobial combinations as a non-thermal, sustainable strategy to improve meat safety. Further research should evaluate the impact of treatments on the sensory attributes of meat and support their implementation at an industrial scale
The future of Scotland’s high streets
No full textThe decline of major retailers has defined a period of turbulence within the UK retail sector, reflecting shifts in consumer behaviour, technology adoption, broader economic distress, and elevated vacancy rates across many urban centres. This has prompted consideration of repurposing retail spaces, with an emphasis on mixed-use developments as a more sustainable alternative to the traditional high street model. To secure strong future for Scotland’s high streets it is required to: (1) promote mixed-use developments: housing, healthcare, education, and leisure; (2) use health and higher education institutions as anchors; (3) promote building reuse strategies; (4) enhance public realm through social and green infrastructure; (5) coordinate national and local funding to address problems not symptoms; and (6) promote stakeholder collaboration
From medical officers of health to multidisciplinary public health specialists: A history of the professional transformation of public health in Britain, 1970 – 2025
No full textThroughout the twentieth century, senior roles in UK public health were reserved for doctors. Local authority medical officers of health were replaced in 1974 by NHS community physicians and from 1989 by medical directors of public health. Over the last decade of the century, an increasingly vocal group of non-medical public health professionals sought to break the glass ceiling that restricted them from advancing to senior roles; although they received encouragement from some leaders within the Faculty of Public Health Medicine, there was also significant resistance from many members. A number of factors came together around the year 2000, which culminated in a ground-breaking decision by the English Department of Health to allow non-medical appointments as directors of public health and consultants in public health in the NHS, with the then Secretary of State memorably declaring it was time to ‘take public health out of the ghetto’. At the same time, the leadership of the Faculty of Public Health Medicine overcame opposition from some of its members and opened its training, examinations, and membership to non-medical candidates. By the early 2020s, half of the renamed Faculty of Public Health members were from backgrounds other than medicine as well as 90% of directors of public health in England. This paper explores the complex history behind this unprecedented opening of a medical specialty to non-medical membership, the factors that enabled it, and the continuing legacy of tensions and inequalities within an occupation that is both a medical specialty and a multidisciplinary profession
Decolonising hydrology: Reflecting on positionalities for sustainable and just futures
No full textColonialism continues to have an enduring impact on hydrological research, practice, and education. This paper examines the colonial legacy in hydrology and highlights the need for decolonisation to achieve justice, inclusivity, and sustainability in water management. Through tracing the development of hydrology research and examining stories from Canada, Indonesia, India, West Asia, and Africa, the paper uncovers the different ways colonial-era policies that have shaped contemporary water challenges. Findings indicate that colonial hydrological practices have led to significant ecological and social impacts, disrupted Indigenous knowledge systems, and intensified transboundary water conflicts. We advocate for considering, when desired by communities, place-based and Indigenous knowledge into the sciences, practices, and outputs of hydrology, revisiting ownership practices, and addressing neo-colonial influences through co-creation and inclusive learning. By reflecting on our colonial past, we propose pathways through reflective questions and a decision tree to guide future water knowledge generation and support local water management
Oil shock spillovers in emerging markets: Sectoral dynamics of demand, supply, and risk channels
No full textThis study investigates how disaggregated oil price shocks (demand, supply, and risk) propagate across emerging market equity sectors. Using quantile granger causality and time-varying parameter vector autoregression extended-joint connectedness frameworks, the analysis captures both dynamic interactions and responses during major turmoil including US-China trade tensions, COVID-19 pandemic, and Russia-Ukraine conflict. Results reveal strong nonlinear and quantile-dependent effects: demand shocks peak under moderate market conditions, supply shocks affect defensive sectors via inflationary channels, and risk shocks exert pervasive systemic influence across all sectors. Connectedness analysis shows high spillovers, with sectors such as financials, industrials, materials acting as major transmitters, while healthcare, utilities, and information technology largely absorb shocks. Network analysis further shows risk shocks as the dominant transmitter during crises, temporarily reversing to a net receiver during COVID-19. These findings highlight crisis-sensitive sectoral vulnerabilities, offering actionable insights for investors and policymakers in oil-sensitive emerging markets
The potential for a transdisciplinary systems approach to improve national policy analysis: Learning from UK cases of home energy transitions
Get PDFThe urgent imperative to decarbonise societies requries effective decisions to neogotiate interconnections of people, technology and policies. In this theory paper, we hypothesise that integrating transdisciplinary engineering with systems approaches can provide useful principles and tools to support effective sustainability policymaking. We consider this hypothesis in the context of a historic and current UK energy sector transition: (a) the transition from ‘town-gas’ to natural gas in the 1960–1970s and (b) the current shift from natural gas to low carbon domestic heating, focussing on heat pump deployment. Through these case-studies, we find that transdisciplinary and systems approaches are apparent in the successful historic transition, while remaining largely absent in the present low carbon heating transition, which is currently stalled. We argue this is caused by policy analysis being siloed and economically focused. We present two systems approach examples to show how they might be applied to begin addressing current UK policy failures for low carbon heating. We identify benefits while recognising some key limitations of this approach, including the resource requirements on officials. The paper concludes with suggestions for further research to continue developing the conceptual and practical basis and therefore lead to improved decision making in national sustainability policymaking