62880 research outputs found
Sort by
Enhancing the robustness of federated learning-based intrusion detection systems in transportation networks
Get PDFAs transportation systems become more connected and automated, strong cybersecurity is essential. This paper presents a novel Intrusion Detection System (IDS) based on Federated Learning (FL) to improve security in transportation networks without compromising user privacy. FL enables multiple devices, such as vehicles and roadside units, to collaboratively train a shared model using local data. We focus on major challenges such as adversarial and data poisoning attacks that can disrupt system operations and reduce detection accuracy. To address these threats, we propose new techniques using model diversity and progressive training. These methods allow different network nodes to train slightly different models, making the system more resilient to attacks. Our experimental results on realworld datasets show that model diversity can improve accuracy by up to 25% under attack scenarios. Progressive training also helps reduce prediction errors by 40%, leading to better overall system performance. The proposed approach strengthens the robustness and adaptability of FL-based IDS in dynamic and hostile transportation environments
The role of soil moisture in the inland penetration of Indian monsoon low-pressure systems
No full textIndian monsoon low-pressure systems (LPSs) are important synoptic-scale cyclonic vortices that typically form over the Bay of Bengal, before penetrating into India where they produce copious precipitation. It has long been argued that high antecedent soil moisture supports their inland penetration. However, recent studies have lent greater support for the role of dynamical processes such as moist barotropic instability. As a result, there is incomplete understanding of the actual role of soil moisture in their inland penetration. In this study, a convection-permitting (3-km horizontal resolution) setup of the Advanced Weather Research and Forecasting (WRF-ARW) model is used to examine the role of soil moisture in the inland penetration of four LPSs, including the well-known case of Gulab–Shaheen during September–October 2021. Soil-moisture perturbations are performed in such a way that either a dry or wet land surface is maintained around LPSs throughout the simulation. Multiple land-surface models (LSMs) are used to test the sensitivity of the results to the choice of LSM. The results suggest that, irrespective of the LSM, the inland penetration of LPSs is not sensitive to soil moisture; although enhancing soil moisture generally increases LPS precipitation when compared with respective control experiments. In contrast, imposed weakening of the magnitude of total surface heat fluxes over land limits the intensification of LPSs and disrupts their tracks, suggesting a greater role of total surface heat fluxes than soil moisture in the inland penetration of LPSs. Thus, this work provides a new perspective on the relationship between soil moisture and LPS propagation
Compromised regeneration, damage to blood vessels and the endomysium underpin permanent muscle damage induced by puff adder (Bitis arietans) venom
No full textThe puff adder (Bitis arietans) is a clinically relevant viper species found throughout Africa,
and it is responsible for a greater incidence of health-related envenomations than all other
snake species on the continent combined. Unresolved skeletal muscle damage is a common
consequence of B. arietans envenomation that can result in long-term morbidity and even
death. Antivenom treatment can mitigate the systemic effects of the venom but offers
little protection against local tissue damage. Identifying the mechanisms through which
B. arietans venom induces tissue damage and impedes skeletal muscle regeneration could
identify possible treatment alternatives that could help alleviate the long-term consequences
of envenomation. Skeletal muscle has an innate ability to regenerate, but constituents
within the venom can impede multiple stages of this regeneration process. In this study, we
employed a combination of biochemical analyses, cell-based assays, and in vivo experiments
to assess the toxicological implications of B. arietans envenomation and its impacts on key
processes of regeneration. Our findings demonstrate that the pathological characteristics of
permanent muscle damage resulting from B. arietans envenomation may be attributed to
the venom’s effects on muscle stem cell precursors, the extracellular matrix (ECM), and the
influence of blood-borne proteins that promote fibrosis
Fox, hedgehog, human, machine: a cautious reappraisal of AI and copyright through rights and welfare
No full textThis chapter examines the challenges that generative AI (GenAI) poses for copyright law and argues for a carefully tailored regulatory approach. Placing the human author at the centre, it contends that fully autonomous AI-generated works, which lack meaningful human creative input, should fall outside the scope of copyright protection. At the same time, in support of a more participatory, socially inclusive, and responsive copyright framework, it argues that where human authors provide sufficiently detailed input and exercise creative choices that shape the final output, such contributions may qualify for protection. This analysis maintains that copyright’s adaptation in the age of GenAI is a cultural opportunity to redefine the future of creativity and advance human development. From a social welfare perspective, such an approach promotes democratic access to creativity and upholds the freedom to create, while highlighting essential questions about inclusivity and the evolving role of creativity in a GenAI-driven society
Breaking away from the pack: uncovering the characteristics of exceptional firms in power-law performance distributions
No full textThis study tracks firm excellence in the wild by exploring the phenomenon of exceptional firm performance, focusing on outliers within a power-law performance distribution. Specifically, we examine organizational, regional, and industry factors contributing to a firm’s exceptional status. Our methodology involves analyzing data on 40,698,121 firm-year observations and 2,696,503 firms in the United Kingdom during 2001–2017 and probing for factors that are associated with firms that achieve exceptional performance. We adopt an abductive reasoning process, guided by previous work on firm acceleration and growth, to explore potential factors influencing such extreme outlier emergence. This flexible approach enables us to delve into predictive characteristics of exceptional firms to help guide policy recommendations. Taken together, what we find expands our understanding of exceptional firm characteristics and sheds light on the organizational contexts where we expect to witness such exceptional performance