17628 research outputs found
Sort by
FedBT: Effective and Robust Federated Unlearning via Bad Teacher Distillation for Secure Internet of Things
Get PDFSmart Internet of Things (IoT) devices generate vast, distributed data, and their limited computational and storage capacities complicate data protection. Federated Learning (FL) enables collaborative model training across clients, enhancing performance and protecting data privacy. The Right to be Forgotten (RTBF) raises the demand for precise data removal. Federated Unlearning (FU) offers a solution for accurate data deletion in FL systems. Existing FU methods often struggle to simultaneously ensure effective data forgetting and preserve model generalization. To mitigate these challenges, an effective and robust FU framework has been proposed, which is based on the “Bad Teacher” knowledge distillation (KD), termed FedBT. First, the “Bad Teacher" KD guides the trained model to eliminate specific client contributions from the global model. Next, the frequency domain extracts the global model’s generalization components. Finally, orthogonal constraints are applied to the KD-generated gradients within the orthogonal subspace of these components, ensuring the gradients preserve the trained model’s generalization ability. FedBT eliminates the need to store historical records of parameter updates. Using orthogonal space constraints, the generalization ability of the trained model is safeguarded during unlearning. Extensive experiments on three datasets with various metrics show our method reduces accuracy by only 0.53% on MNIST, 0.26% on Fashion-MNIST, and 4.67% on CIFAR10, surpassing the best approach. Furthermore, FedBT obtains an unlearning performance that most closely approximates the results obtained from retraining from scratch. FedBT boosts IoT security by enabling the “forgetting" of certain client data, crucial for protecting user privacy and ensuring secure device interactions
Experiences of Social Disconnection in a Bereaved Community Sample from Ontario, Canada
Get PDFLower perceived social support is a known risk factor for problematic grief reactions, but specific facets such as social disconnection may play a critical role in shaping grief responses. This study utilized the Oxford-Grief Social Disconnection Scale (OG-SD) to examine the demographic, loss-related, and psychological correlates of its three core dimensions, as identified by Smith et al. (2020): Negative Interpretation of Others’ Reactions to Grief Expression, Altered Social Self, and Safety in Solitude. Participants were a non-probability sample of N =1171 bereaved adults living in Ontario, Canada. Confirmatory factor analysis (CFA) was used to confirm the three dimensions of grief-related social disconnection identified by Smith et al. (2020). Correlation and one-way ANOVA tests explored demographic and loss-related correlates of these dimensions, while associations with symptoms of Prolonged Grief Disorder (PGD), depression, and anxiety were assessed through correlational analyses. CFA results confirmed that the OG-SD was best reflected by a correlated three-factor model comprising Negative Interpretation of Others’ Reactions to Grief Expression, Altered Social Self, and Safety in Solitude latent variables. Distinct associations between the core dimensions of social disconnection and loss-related variables were identified, and significant associations between all three dimensions and scores on measures of PGD, depression, and anxiety were also observed. Findings from this study not only provide additional support for the validity and reliability of the OG-SD in a general population sample of Canadian adults, but also, for the first time, identify demographic, loss-related, and psychological factors associated with social disconnection
A Novel TLS-Based Fingerprinting Approach That Combines Feature Expansion and Similarity Mapping
Get PDFMalicious domains are part of the landscape of the internet but are becoming more prevalent and more dangerous both to companies and to individuals. They can be hosted on various technologies and serve an array of content, including malware, command and control and complex phishing sites that are designed to deceive and expose. Tracking, blocking and detecting such domains is complex, and very often it involves complex allowlist or denylist management or SIEM integration with open-source TLS fingerprinting techniques. Many fingerprinting techniques, such as JARM and JA3, are used by threat hunters to determine domain classification, but with the increase in TLS similarity, particularly in CDNs, they are becoming less useful. The aim of this paper was to adapt and evolve open-source TLS fingerprinting techniques with increased features to enhance granularity and to produce a similarity-mapping system that would enable the tracking and detection of previously unknown malicious domains. This was achieved by enriching TLS fingerprints with HTTP header data and producing a fine-grain similarity visualisation that represented high-dimensional data using MinHash and Locality-Sensitive Hashing. Influence was taken from the chemistry domain, where the problem of high-dimensional similarity in chemical fingerprints is often encountered. An enriched fingerprint was produced, which was then visualised across three separate datasets. The results were analysed and evaluated, with 67 previously unknown malicious domains being detected based on their similarity to known malicious domains and nothing else. The similarity-mapping technique produced demonstrates definite promise in the arena of early detection of malware and phishing domains
Circulating structural timber and engineered wood products – challenges and potentials towards reliable evaluation of mechanical properties
Get PDFCirculating structural timber and engineered wood products require reliable assessment of key mechanical properties. However, existing standards for strength grading sawn timber are not designed for the reuse and recycling of timber and are unsuitable for this in a number of ways. Furthermore, existing procedures for the re-assessment of structural components are mainly focused on the identification of in-use damage and assumptions about the original mechanical properties, and not on the quantitative re-evaluation of the current mechanical properties. In this paper, existing strength grading procedures in Europe are discussed according to their potential for the assessment of reused timber, with observations on shortcomings of the underlying basis when not applied to new, not previously graded timber. Since the situation for reused and recycled laminated components (glued laminated timber and cross-laminated timber) might be simpler, and perhaps more commercially relevant, a framework will be presented to estimate the mechanical properties of these structural components based on the load history and non-destructive assessment methods. The basis could be expanded by future work to allow the re-grading of sawn timber
Inclusive Kerbs Study Phase 3
No full textThe Inclusive Kerbs Phase 3 research project gathered data from a representative example of kerbs within The City of Edinburgh to understand what kerbs are present in the city and how people with various impairments use them. The study was conducted by Mott MacDonald and Edinburgh Napier University’s Transport Research Institute. It was commissioned by Transport Scotland on behalf of the Scottish Road Research Board (SRRB) and the Department for Transport (DfT).Phase 1 of the project looked at existing research and found that there are few studies on inclusive kerbs considering both engineering and human factors.Phase 2 gathered data and tried out selected methods to learn more about how people use kerbs. The study considered how kerbs are used for navigating along a street and for crossing the street.Phase 3 collected data from eleven people with severe to moderate impairments through further online interviews and site visits with volunteer participants. The information gathered from the interviews and site visits are compared against the survey data to identify any patterns in experience
Nonlinear Shear Waves in Compressible Media: Occurrence of Strong Shocks
Get PDFApparently for the first time, shear shock wave fronts (shear shocks) are observed in a hyperfoam at the propagation of shear waves. The hyperfoam is modelled by the Ogden compressible hyperelastic potential. A possible appearance of the shear shocks may explain the kinetic and strain energy attenuation along with heat release at the propagation of shear waves in hyperfoams. The analysis is based on the Cauchy formalism for equations of motion, equations of energy balance, and FE analysis for solutions of the constructed nonlinear hyperbolic equation
Dynamic Event-Triggered Sliding-Mode Bipartite Consensus for Multi-Agent Systems With Unknown Dynamics
No full textThis paper addresses a data-driven sliding mode bipartite consensus issue for nonlinear discrete-time multi-agent systems with antagonistic interactions and limited communication resources. Initially, the signed graph theory is employed, and a combined measurement error function is formulated, transforming the bipartite consensus issue into a traditional consensus issue. An enhanced compact form dynamic linearization model is then established based on the input/output data and the formulated combined measurement error function. Moreover, a dynamic event-triggered function and a sliding-mode surface are designed, leading to the development of a fully distributed dynamic event-triggered sliding-mode bipartite consensus (DET-SMBC) approach. The proposed DET-SMBC approach is subsequently extended to a fully distributed dynamic event-triggered robust sliding-mode bipartite consensus (DET-RSMBC) scheme to improve robustness. The convergences of the tracking errors of both methods are rigorously deduced. Finally, simulation studies and hardware experiments are conducted to demonstrate the effectiveness of the proposed methods. Note to Practitioners—In multi-agent systems, the applicability of existing methods can be reduced by some issues, such as uncertain dynamics models, unknown disturbances, and the limitation of communication bandwidth. These issues can influence existing methods’ usefulness and cause instability, so DET-SMBC and DET-RSMBC methods are proposed in this paper. Compared with existing results, identifying a precise dynamics model for each controlled plant is unnecessary, the necessity of high-performance hardware for data transmission is relieved, and the effects of unknown disturbances are reduced. Moreover, the proposed methods are applied to realistic servo motor systems to conduct speed bipartite consensus tasks well. It is noted that most complicated mechanisms are controlled by servo motors, so the proposed methods can be applied to more practical engineering systems
Air pollution-associated chronic kidney disease (APA-CKD): evidence from a cross-sectional study of Niger Delta communities
Get PDFObjective: Air pollution is an emerging risk factor for chronic kidney disease (CKD) that is typically ignored in preventive interventions. This study investigated whether long-term exposure to ambient air pollution in communities near petrochemical industries in the Niger Delta was associated with CKD. Design: A cross-sectional study with an embedded citizen science inquiry. Settings: Four communities situated at varying distances from a petrochemical refinery in Niger Delta, Nigeria. Participants: We obtained sociodemographic, behavioural, exposure history and clinical data from 1460 participants who have resided for at least 5 years in the four communities. A citizen science approach was used to monitor air pollutant concentrations with eight community volunteers. Results: The mean PM2.5, PM10 and volatile organic compounds (VOC) concentrations exceeded the WHO-acceptable limits in all four communities. CO2 was acceptable in the farthest communities from the refinery, while O3 was within acceptable limits in all communities. The total hazard quotient was relatively higher in the two communities near the refinery (11.27, 11.63) than those farther (9.63, 10.68), F=0.038, p=0.989. The overall prevalence of CKD was 12.3%; it was 17.9% in the community closest to the refinery and 8.0% in the farthest (χ2=18.292, p=0.004). Increasing age was the only independent risk factor for CKD after adjusting for confounding factors and intrahousehold design effect (adjusted OR 1.26; 95% CI 1.09 to 1.45, p=0.002). Conclusion: Long-term exposure to ambient air pollution may increase CKD risk in susceptible populations. Social factors and environmental exposures associated with CKD are prevalent in the communities, necessitating multifaceted and inclusive approaches to mitigate air pollution and the associated kidney disease risks. More studies are required to explore the mechanism of air pollution-associated kidney disease and interventions to reverse or limit it
AI-Enhanced High-Speed Data Encryption System for Unmanned Aerial Vehicles in Fire Detection Applications
Get PDFSmall unmanned aerial vehicles (UAVs) are increasingly used for wildfire detection, where they must not only identify fire events rapidly but also transmit large volumes of sensor data securely to ground stations. Achieving both fast on-board analysis and high-speed encrypted data transmission within the size, weight, and power limits of UAV platforms remain a major technical challenge. In this study, we introduce a compact, FPGA-based system that simultaneously performs real-time fire detection and high-throughput data encryption. Our system integrates a programmable logic chip (FPGA), deep-learning models for visual recognition, and AES-256 cryptographic cores onto a single hardware module. A key innovation is a shared scheduling mechanism that coordinates these two functions efficiently. Furthermore, we demonstrate how artificial intelligence contributes beyond image classification: a lightweight neural network monitors input data streams and dynamically adjusts encryption key parameters, thereby improving security without compromising performance. The hardware supports encrypted data transfer rates of 800 megabits per second at a latency of just 2 microseconds, while identifying fire signatures at 30 frames per second. Extensive testing, including cross-validation on a 50,000-frame dataset and environmental stress testing from-20 °C to 55 °C, confirms robust performance under real-world conditions. While the current memory footprint limits multi-camera input, this work offers a foundational design for future systems that aim to combine edge computing, secure communications, and AI-driven perception in autonomous aerial platforms
Data sources:
Access Repository Dashboard
Do you manage Repository@Napier? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!