260851 research outputs found
Sort by
FedDiverse: tackling data heterogeneity in federated learning with diversity-driven client selection
No full textFederated Learning (FL) enables decentralized training of machine learning models on distributed data while preserving privacy. However, in real-world FL settings, client data is often non-identically distributed and imbalanced, resulting in statistical data heterogeneity which impacts the generalization capabilities of the server's model across clients, slows convergence and reduces performance. In this paper, we address this challenge by proposing first a characterization of statistical data heterogeneity by means of 6 metrics of global and client attribute imbalance, class imbalance, and spurious correlations. Next, we create and share 7 computer vision datasets for binary and multiclass image classification tasks in Federated Learning that cover a broad range of statistical data heterogeneity and hence simulate real-world situations. Finally, we propose FedDiverse, a novel client selection algorithm in FL which is designed to manage and leverage data heterogeneity across clients by promoting collaboration between clients with complementary data distributions. Experiments on the seven proposed FL datasets demonstrate FedDiverse's effectiveness in enhancing the performance and robustness of a variety of FL methods while having low communication and computational overhead.</p
[Abstract] Hospitalisation with COVID 19 is associated with worse vascular remodelling in men: The CARTESIAN study
No full textObjective: COVID-19 can lead to a persistent low-grade inflammatory that could be potentially associated with arterial remodelling and potential atherosclerotic changes, resulting in an increased risk of cardiovascular disease. The aim of this study was to evaluate sub-clinical atherosclerosis and arterial remodelling in patients post COVID-19 infection.Methods: This is a prospective, multi-centric, cohort, including 15 countries worldwide. Participants were recruited 6 months after COVID19 infection. Carotid intima-media thickness (IMT) was evaluated by processing longitudinal B-mode ultrasound scans (≥7.5 MHz linear array transducer). Left and right carotid arteries were imaged 1-2cm proximal to the carotid bifurcation and the mean of the two values was used for the analysis. Participants were classified into 4 groups-COVID-19 negative control (NC) (n = 107), and three COVID-19 positive groups classified by severity: not hospitalized (or hospitalization for less than 24 hours) (NHP) (n = 270), hospitalized patients in medical unit (HP) (n = 246), and patients requiring ICU/resuscitation care (ICU) (n = 65).Results: Patients in the HP and ICU groups were significantly older compared to controls NC 44(28-58), NHP 42(31-52), HP 57(49-66), ICU 58(51-66) years. Both unadjusted/adjusted hierarchical linear models (for age, sex, BMI, smoking status, established CVD, diabetes, and hypertension) showed a significantly higher IMT in HP compared with NC (adjusted mean (95% CI): 0.745 [0.69–0.80] mm vs 0.66 (0.60–0.72) mm). In addition, a significant group × sex interaction was observed, with sex-specific analyses showing an approximately twofold greater increase in IMT in HP men compared with HP women. Wall to lumen ratio was also higher in HP (0.33(0.30;0.37)) vs NC (0.26(0.22;0.30)), with a significant group × sex interaction. After adjusting for confounders, wall lumen ratio in HP compared to NC was twice as higher in men in comparison to women.Conclusions: Patients hospitalised for more than 24 hours during COVID 19 show impaired vascular characteristics, with a greater disease-associated vascular remodelling in men compared to women. This could be potentially due to a combined effect of pre-existing vascular damage that predispose them to severe COVID-19 and accelerated vascular remodelling due to persistent chronic inflammation.</p
Design, experimental testing and optimization of a high-precision linear delta robotic platform for microsurgery
No full textTraditional surgical methods and existing robotic platforms have made great successes in general surgery field. These approaches have significantly improved surgical outcomes by enhancing dexterity, reducing invasiveness, and enabling more consistent procedural performance. However, their design philosophy—primarily based on the requirements of normal-scale surgical procedures—introduces limitations when applied to more delicate operations. These include constraints in precision, stability, and structure complexity. This gap is most apparent in microsurgery, which demands the manipulation of structures on the scale of just a few micrometers, requiring submillimeter-level precision and exceptional stability. While surgical microscopes are essential for visualization in such procedures, their use still introduces practical challenges, including limited instrument maneuverability, and ergonomic strain during prolonged operations, thereby motivating the need for robotic assistance to enhance precision and operational efficiency. As a result, the demand for high-precision surgical robots has been growing in the medical field, particularly for procedures such as neurosurgery, plastic surgery, ophthalmic microsurgery, etc.To address these challenges, several robotic platforms have been developed for microsurgery. Systems such as the MUSA system and the MMI Symani Surgical System have demonstrated significant advancements in enhancing surgical precision, primarily through motion scaling, tremor filtering, and teleoperated control architectures that improve motion stability and accuracy beyond human physiological limits. However, these platforms still present key limitations, including high costs, complex operation procedures, and steep learning curves for both surgeons and technical staff, and structurally complex multi-arm architectures that increase system footprint and integration difficulty in constrained surgical environments. These limitations highlight an urgent need for a compact, cost-effective, and structurally efficient robotic platform that enables high-precision manipulation while reducing mechanical complexity and facilitating integration into constrained microsurgical environments.To address these needs, this thesis presents the design and development of a cost-effective, high-precision robotic platform based on a linear delta mechanism. The proposed system is capable of achieving submillimeter-level resolution, tailored for microsurgical procedures. The robot consists of three actuators driven by stepper motors, combined with a microscope-based visual navigation system. The control software, developed in LabVIEW, enables precise motor control and real-time trajectory execution using PWM signals. The robot’s performance was evaluated through simulations and experimental tests. The integration of a closed-loop control system was also developed and tested to minimize positional errors and improve the overall trajectory performance, which achieves a final precision of 2.6 ± 0.2 μm in the X-Y plane, and improved stability in controlled stepping modes. The results also showed that under closed-loop control, trajectory repeatability and angular accuracy improved by 23.4% and 26.8%, respectively, compared to the open-loop baseline. The system shows significant promise for high-precision microsurgical applications, including plastic surgery, ophthalmology and neurosurgery, while maintaining cost-effectiveness for broader adoption.</p
Research Data for Absorption Imaging of Quantum Gases Near Surfaces Using Incoherent Light
No full textThe dataset is used in the paper published on ArXiv. It is a collection of measurement data such as raw images, optical density, atomic column density or integrated atomic density data, obtained by analysing raw images, and current calibration data. All data is in standard data formats such as png and txt. From the dataset all figures can be reproduced, and the folder structure and naming convention helps to identify each file for this purpose. The folders contain readme.txt files for further guidance.Abstract: We introduce an absorption imaging technique for ultracold gases that suppresses interference fringes and coherence-induced artifacts by reducing the transverse spatial coherence of the imaging light. The method preserves the narrow spectral bandwidth required for resonant absorption imaging and is implemented as a modular extension to standard imaging setups using a rotating diffuser. We demonstrate tunability of the illumination light's coherence without modifying the imaging optics. Using this approach, we achieve reliable imaging of ultracold atomic clouds in micron-scale proximity to complex surfaces, where standing waves, edge diffraction, and speckle severely limit conventional absorption imaging.</p
Data for paper: Reductive Cyclopentamerization of Carbon Monoxide to an Elusive Croconate Radical Tri-Anion
No full textData for paper published in Journal of the American Chemical Society (March 2026)CIF files for all compounds.checkcif files for all compounds.EPR spectroscopy data.UV/vis spectroscopy data for all compounds.Raman spectroscopy data.Magnetic measurements.AbstractCoupling reactions of carbon monoxide represent an ideal synthetic route to cyclic oxocarbons. Whereas oxocarbon anions based on rings of three, four and six carbon atoms have been synthesized from CO coupling, cyclic five-membered croconate anions have evaded capture by this method. Here, we show that CO coupling initiated by rare-earth dinitrogen complexes in concert with molybdenum hexacarbonyl yields rare-earth complexes of the exotic croconate radical trianion, [C₅O₅]³⁻, a species only observed previously under cryogenic irradiation. Contrasting reactions of the rare-earth dinitrogen complexes with CO alone produce ketene-carboxylate anions, highlighting the crucial templating role of molybdenum in forming croconate. Spectroscopic and computational analyses establish the radical nature of the croconate anion and reveal a heterobimetallic reaction pathway for its assembly with multi-electron transfer. This discovery surmounts a long-standing challenge in oxocarbon chemistry, illustrating how strongly reducing rare-earth compounds and mixed-metal cooperativity can enable the isolation of sought-after organic radicals.</p
Support, Care, and Mutual Aid in Protracted Displacement Economies
No full textIn settings of protracted displacement, external sources of material support typically dry-up when international donors lose interest in any given conflict and ensuing displacement crisis. Populations in those settings must rely on different forms of support, financial, and non-financial, with some reciprocal or mutual in their nature. The aim of this paper is to show how displacement affected populations provide and receive different forms of support, and the importance of non-financial acts of care and support in protracted displacement economies. It also considers mutual aid by discussing the reciprocal features of this support. To do so, we present some qualitative and quantitative empirical findings from a major three-year-long research project that spanned across five countries, with four discussed here, and three sites in each of those countries. The extensive original dataset consists of over 15,000 household surveys, 682 qualitative interviews, 60 focus groups and stakeholder workshops, as well as five film-making workshops. The paper presents research findings from Ethiopia, Lebanon, Myanmar, and Pakistan. </p
National social cost of carbon: an application of FUND
No full textThis paper presents a refined country-level integrated assessment model, Framework for Uncertainty, Negotiation, and Distribution (FUND) 3.9n, that extends the regional FUND 3.9 framework by incorporating sector-specific climate impact functions and parametric uncertainty analysis for 198 individual countries. The model enables estimation of the national social cost of carbon (NSCC), capturing heterogeneity across nations from economic structure, climate sensitivity, and population exposure. Our results demonstrate that both the NSCC and the global sum estimates are highly sensitive to damage specifications and preference parameters, including the pure rate of time preference and relative risk aversion. Compared to aggregated single-sector approaches, the disaggregated model with uncertainty yields higher values of the NSCC for low- and middle-income countries. The paper contributes to the literature by quantifying how sector-specific vulnerabilities and stochastic variability amplify climate damages and reshape global equity in the distribution of the NSCC. The NSCCs derived from our model offer policy-relevant metrics for adaptation planning, mitigation target setting, and equitable burden-sharing in international climate negotiations. This approach bridges the gap between globally harmonized carbon pricing and nationally differentiated climate impacts, providing a theoretically grounded and empirically rich framework for future climate policy design.</p
The Co-Enactment of the Leader-Follower Identity
No full textIn this chapter the authors start from the recognition that most people in organizations are both a leader and a follower, and that theory is lacking on how these identities are constructed and enacted together. The authors therefore develop theory on how the leader-follower identity is co-enacted. That is to say, they conceptualize how one individual constructs their leader-follower identity and behaves according to that construction. The authors use metaphors to conceptualize four forms of co-enactment, along two dimensions. The first of these dimensions regards whether leader-follower identities are conceptualized as fixed to predefined roles, or fluid; the second, on the relationships between the two identities, which can be seen as dissonant or resonant. From their theoretical exploration they extrapolate four typologies, which they define through metaphorical figures: the switcher (when identities are seen as fixed and dissonant); the negotiator (fixed, resonant); the juggler (fluid, resonant); and the struggler (fluid, dissonant). The authors briefly discuss when and how these typologies of co-enactments are more likely to take place.</p
Phase-averaged stresses and elastic load partitioning in an AL-8(Ce, La) alloy measured by in-situ synchrotron X-ray diffraction under uniaxial tensile testing
No full textAl-(Ce, La) alloys are potential alternatives to Al-Si alloys for automotive and aerospace applications, offering strength retention up to 300 °C due to the excellent thermal stability of Al₁₁(Ce, La)₃ as the primary strengthening phase. However, they show limited strength at room temperature and the origin of this limitation is poorly understood. In particular, it is unclear how the applied macroscopic stress is shared between the Al matrix and the Al₁₁(Ce, La)₃ phase, and how to increase the load sharing of (Ce, La)-containing phases without compromising ductility or high-temperature stability. This study investigated the load partitioning between the primary aluminium crystals and the Al₁₁(Ce, La)₃ intermetallic particles in an Al-8(Ce, La) alloy during uniaxial tensile testing. In-situ synchrotron X-ray diffraction was used to monitor the evolution of lattice strains, allowing the calculation of stress carried by the Al₁₁(Ce, La)₃. This methodology can now be used to investigate compositional and microstructural changes to the strengthening phase Al₁₁(Ce, La)₃, allowing modification effectiveness to be compared and quantified consistently.</p
Synergistic motifs in Gaussian systems
No full textHigh-order interdependencies are central features of complex systems, yet a mechanistic explanation for their emergence remains elusive. Currently, it is unknown under what conditions high-order interdependencies, quantified by the information-theoretic construct of synergy, arise in systems governed by pairwise interactions. We solve this problem by providing precise sufficient and necessary conditions for when synergy prevails over low-order interdependencies in the weak interaction regime, namely, we prove that antibalanced (highly frustrated) correlational structures in Gaussian systems are sufficient for synergy-dominance and that antibalanced interaction motifs in Ornstein–Uhlenbeck processes are necessary for synergy-dominance. We validate the applicability of these analytical insights beyond the weak interaction regime, as well as in Ising, oscillatory, and empirical networks from multiple domains. Our results demonstrate that pairwise interactions can give rise to synergistic information in the absence of explicit high-order mechanisms, and highlight structural balance theory as an instrumental conceptual framework to study high-order interdependencies.</p