11563 research outputs found
Sort by
LoRA-FL: a low-rank adversarial attack for compromising group fairness in federated learning
No full textFederated Learning (FL) enables collaborative model training without sharing raw data, but agent distributions can induce unfair outcomes across sensitive groups. Existing fairness attacks often degrade accuracy or are blocked by robust aggregators like KRUM. We propose LoRA-FL: a stealthy adversarial attack that uses low-rank adapters to inject bias while closely mimicking benign updates. By operating in a compact parameter subspace, LoRA-FL evades standard defenses without harming accuracy. On standard fairness benchmarks (Adult, Bank, Dutch), LoRA-FL reduces fairness metrics (DP, EO) by over 40% with only 10�20% adversarial agents, revealing a critical vulnerability in FL�s fairness-security landscape. Our code base is available at: https://github.com/ sankarshandamle/LoRA-FL
Numerical simulation of Airfoil-Vortex Interaction in Tandem Airfoils
No full textAerodynamic noise generation from aircraft and wind turbines is one of the pertinent problems in the aeronautical industry. The fluid flow over an airfoil generates noise because of various phenomena such as vortex-shedding and flow separation. The vortex shedding from the upstream object interacts with the object placed in the wake and generates noise, called body-vortex interactions (BVI). Tandem airfoil arrangements were used to analyze the BVI effect on noise generation. The hybrid RANS-LES method was used for simulation. The noise generated in the two airfoils case showed a slight increase. The acoustic level depended on the interference pattern between the acoustic waves radiated from the trailing edge of both airfoils and the interaction of the shed vortices with the second airfoil. It was found that the flow pattern and the resultant acoustic field depend on the distance between the airfoils. An increase in the inlet velocity shifted the peak SPL amplitudes to higher frequencies. The modal analysis provided insight into the broadband spectrum. The trailing edge of the airfoil was observed to generate pressure waves of a broadband nature containing multiple frequencies. Different dominating modes were found in the two airfoil cases, which resulted in a modified acoustic field. � 2024 Elsevier B.V., All rights reserved
Thermal analysis of a building in hot and dry climate: a detailed study
No full textActive cooling is quickly becoming a necessity in hot-dry climates prevailing across large sections of the world, where the equivalent solar temperature can rise above 70 °C in the summer. The cooling demand, within the structure depends on various components of building, some of which may be directly exposed to sunlight. While a few mitigating technologies that reduce this energy ingress have been proposed, quantitative analyses leading to policy prescriptions are largely missing. This research is aimed at identifying the building elements primarily responsible for the energy ingress by focusing on a specific apartment located in a hot and dry climate (in western India). A comprehensive resistance–capacitance (RC) model inclusive of all of the building components has been used for this purpose. Among other insights, this model shows that while a heat ingress largely occurs through the roof, the installation of the Brick Bat Koba waterproofing and sand bedding and reflective tiles lead to a reduction in energy ingress. However, these two mitigating strategies are not adequate for achieving thermal comfort during summer and the presence of brick bat Koba and sand bedding worsens the situation by preventing the indoor environment from cooling down sufficiently during the night. As potential solutions, we propose the usage of nocturnal ventilation using fresh ambient air, which helps to reduce the indoor temperature during the night, concomitantly reducing the energy consumption attributable to air-conditioning. These insights are expected to lead to the adoption of best practises and ultimately to the development of regulatory guidelines
Programmable DNA-Based Nanodevices for Next-Generation Clinical and Healthcare Applications
No full textDeoxyribonucleic acid (DNA) nanotechnology has brought an unparalleled set of possibilities for self-assembled structures emerging as an independent branch of synthetic biology. The field of science uses the molecular properties of DNA to build nanoparticles and nanodevices that have the potential to bring breakthroughs in medical science. On the one hand, their biocompatibility, precision, synthetic ease, and programmability make them an ideal choice in drug delivery and healthcare. On the other, the lack of proper biodistribution profiles, stability inside the system, enzymatic cleavage, immune recognition, and translational barriers are some of the hurdles it faces. Many recent technological advancements are in progress to tackle these challenges, while some already have been used. These tools and technologies need to be understood and studied for the successful transition of these intelligent DNA nanostructures (DNs) to healthcare applications. This review thus, highlights some of the challenges being faced by the DNs in healthcare. Additionally, it provides an overview of the recent trends in using these devices in disease detection and remission and finally talks about the future scope and opportunities for an effective transition from bench to bedside
Impacts of irrigation expansion on moist-heat stress based on IRRMIP results
No full textIrrigation rapidly expanded during the 20th century, affecting climate via water, energy, and biogeochemical changes. Previous assessments of these effects predominantly relied on a single Earth System Model, and therefore suffered from structural model uncertainties. Here we quantify the impacts of historical irrigation expansion on climate by analysing simulation results from six Earth system models participating in the Irrigation Model Intercomparison Project (IRRMIP). Results show that irrigation expansion causes a rapid increase in irrigation water withdrawal, which leads to less frequent 2-meter air temperature heat extremes across heavily irrigated areas (≥4 times less likely). However, due to the irrigation-induced increase in air humidity, the cooling effect of irrigation expansion on moist-heat stress is less pronounced or even reversed, depending on the heat stress metric. In summary, this study indicates that irrigation deployment is not an efficient adaptation measure to escalating human heat stress under climate change, calling for carefully dealing with the increased exposure of local people to moist-heat stress
L3D-Pose: Lifting Pose for 3D Avatars from a Single Camera in the Wild
No full textWhile 2D pose estimation has advanced our ability to interpret body movements in animals and primates, it is limited by the lack of depth information, constraining its application range. 3D pose estimation provides a more comprehensive solution by incorporating spatial depth, yet creating extensive 3D pose datasets for animals is challenging due to their dynamic and unpredictable behaviours in natural settings. To address this, we propose a hybrid approach that utilizes rigged avatars and the pipeline to generate synthetic datasets to acquire the necessary 3D annotations for training. Our method introduces a simple attention-based MLP network for converting 2D poses to 3D, designed to be independent of the input image to ensure scalability for poses in natural environments. Additionally, we identify that existing anatomical keypoint detectors are insufficient for accurate pose retargeting onto arbitrary avatars. To overcome this, we present a lookup table based on a deep pose estimation method using a synthetic collection of diverse actions rigged avatars perform. Our experiments demonstrate the effectiveness and efficiency of this lookup table-based retargeting approach. Overall, we propose a comprehensive framework with systematically synthesized datasets for lifting poses from 2D to 3D and then utilize this to re-target motion from wild settings onto arbitrary avatars. The L3D-Pose dataset can be found at https://soumyaratnadebnath.github.io/L3D-Pose
The sky remembers everything: celestial amplitude, shadow and OPE in quadratic EFT of gravity
No full textIn this paper, we compute the celestial amplitude arising from higher curvature corrections to Einstein gravity, incorporating phase dressing. The inclusion of such corrections leads to effective modifications of the theory's ultraviolet (UV) behaviour. In the eikonal limit, we find that, in contrast to Einstein's gravity, where the and -channel contributions cancel, these contributions remain non-vanishing in the presence of higher curvature terms. We examine the analytic structure of the resulting amplitude and derive a dispersion relation for the phase-dressed eikonal amplitude in quadratic gravity. Furthermore, we investigate the celestial conformal block expansion of the Mellin-transformed conformal shadow amplitude within the framework of celestial conformal field theory (CCFT). As a consequence, we compute the corresponding operator product expansion (OPE) coefficients using the Burchnall-Chaundy expansion. In addition, we evaluate the OPE via the Euclidean OPE inversion formula across various kinematic channels and comment on its applicability and implications. Finally, we briefly explore the Carrollian amplitude associated with the corresponding quadratic EFT