159370 research outputs found
Sort by
Reasonable Doubt in the Face of Bias:Fair Flagging with Dirichlet-Based Models
No full textEnsuring safety in machine learning requires not only robustness to adversarial or distributional uncertainty but also protection against systematic bias. Models that produce unfair or group dependent predictions pose critical risks when deployed in socially sensitive domains such as credit, justice, or healthcare. This work introduces EviFair, a fairness-aware safety monitor that flags predictions exhibiting excessive dependence on protected attributes. EviFair combines evidential uncertainty modelling with sensitivity analysis to detect biased decision paths, flagging predictions where fairness cannot be guaranteed. We also show that EviFair’s bias scores can effectively guide post-processing fairness methods. Results on standard fairness benchmark datasets show that EviFair achieves substantial reductions in group disparities with minimal impact on predictive performance, demonstrating its promise as a practical, inference-time mechanism for bias-sensitive, safety-aware model oversight
A critical comparison of polypropylene and polyurethane sling materials after implantation in a suburethral sheep model
No full textAlthough polypropylene (PP) materials have been implanted for decades for urethral support in the pelvic floor, appropriate large animal models and advanced materials analysis techniques have not previously been used to investigate the clinical problems they can cause - inflammation, pain and erosion through tissues An ovine model duplicating the surgical procedure for suburethral sling surgery was developed. Here we present the results after 3 months implantation using immunohistochemistry and advanced materials characterisation of two materials PP and Polyurethane (PU). Both materials were well integrated into the tissue. The M1/M2 ratio in PP-implanted tissue was statistically significantly elevated (4.29) compared to PU (0.63) and control tissue (0.34). The higher ratio indicates a more inflammatory response to PP than PU. Surface roughness (assessed using atomic force microscopy) increased in both materials, Rq from 5.73-10.2nm in PP and from 1.03-2.96nm in PU; whilst Ra went from 4.75-7.85nm in PP and from 0.81-2.36nm in PU. Notably, surface stiffness increased by 0.05GPa in PP and decreased by 0.2GPa in PU. PP underwent both surface and bulk material degradation, PU did not. Detailed testing of implantable materials in an appropriate animal model should be conducted before materials are introduced into clinical practice. It is salutary that this has never been reported before. The use of material characterisation techniques allowed us to identify problems in the performance of PP, notably surface degradation, changes in bulk properties and stiffening, which can activate macrophages. In contrast, PU appears a more suitable alternative material for use in treating patients with SUI
Politicised Changes to the NICE Threshold Risk Making Cost-Effectiveness Analysis Performative, Not Informative
Get PDFThe global spectra-trait initiative: a database of paired leaf spectroscopy and functional traits associated with leaf photosynthetic capacity
Get PDFAccurate assessment of leaf functional traits is crucial for a diverse range of applications from crop phenotyping to parameterizing global climate models. Leaf reflectance spectroscopy offers a promising avenue to advance ecological and agricultural research by complementing traditional, time-consuming gas exchange measurements. However, the development of robust hyperspectral models for predicting leaf photosynthetic capacity and associated traits from reflectance data has been hindered by limited data availability across species and environments. Here we introduce the Global Spectra-Trait Initiative (GSTI), a collaborative repository of paired leaf hyperspectral and gas exchange measurements from diverse ecosystems. The GSTI repository currently encompasses over 7500 observations from 397 species and 41 sites gathered from 36 published and unpublished studies, thereby offering a key resource for developing and validating hyperspectral models of leaf photosynthetic capacity. The GSTI database is developed on GitHub (https://github.com/plantphys/gsti, last access: 4 January 2026) and published to ESS-DIVE https://doi.org/10.15485/2530733, Lamour et al., 2025). It includes gas exchange data, derived photosynthetic parameters, and key leaf traits often associated with traditional gas exchange measurements such as leaf mass per area and leaf elemental composition. By providing a standardized repository for data sharing and analysis, we present a critical step towards creating hyperspectral models for predicting photosynthetic traits and associated leaf traits for terrestrial plants
A solution to the S8 tension through neutrino–dark matter interactions
Get PDFNeutrinos and dark matter (DM) are two of the least understood
components of the Universe, yet both play crucial roles in cosmic evolution.
Clues about their fundamental properties may emerge from discrepancies
in cosmological measurements across different epochs of cosmic history.
Possible interactions between them could leave distinctive imprints on
cosmological observables, offering a rare window into dark sector physics
beyond the standard ΛCDM framework. Here we present compelling
evidence that DM–neutrino interactions can resolve the persistent structure
growth parameter discrepancy, S8 = σ8 √Ωm/0.3, between early and late
Universe observations. By incorporating cosmic shear measurements from
current weak lensing surveys, we demonstrate that an interaction strength
of u ≈ 10−4 not only provides a coherent explanation for the high-multipole
observations from the Atacama Cosmology Telescope, but also alleviates
the S8 discrepancy. Combining early Universe constraints with DES Y3
cosmic shear data yields a nearly 3σ preference for non-zero DM–neutrino
interactions. This strengthens previous observational claims and provides a
clear path towards a breakthrough in cosmological research. Our findings
challenge the standard ΛCDM paradigm and highlight the potential of future
large-scale structure surveys, which can rigorously test this interaction and
unveil the fundamental properties of DM
Mid-infrared InAs/InP quantum-dot lasers
Get PDFMid-infrared semiconductor lasers operating in the 2.0–5.0 μm spectral range play an important role for various applications, including trace-gas detection, biomedical analysis, and free-space optical communication. InP-based quantum-well (QW) and quantum-dash (Qdash) lasers are promising alternatives to conventional GaSb-based QW lasers because of their lower cost and mature fabrication infrastructure. However, they suffer from high threshold current density (Jth) and limited operation temperatures. InAs/InP quantum-dot (QD) lasers theoretically offer lower Jth owing to their three-dimensional carrier confinement. Nevertheless, achieving high-density, uniform InAs/InP QDs with sufficient gain for lasing over 2 μm remains a major challenge. Here, we report the first demonstration of mid-infrared InAs/InP QD lasers emitting beyond 2 μm. Five-stack InAs/In0.532Ga0.468As/InP QDs grown by molecular-beam epitaxy exhibit room-temperature photoluminescence at 2.04 μm. Edge-emitting lasers achieve lasing at 2.018 μm with a low Jth of 589 A cm−2 and a maximum operation temperature of 50 °C. Notably, the Jth per layer (118 A cm−2) is the lowest ever reported for room-temperature InP-based mid-infrared lasers, outperforming QW/Qdash counterparts. These results pave the way for a new class of low-cost, high-performance mid-infrared light sources using InAs/InP QDs, marking a notable step forward in the development of mid-infrared semiconductor lasers
Does data drive campaign decision-making? Theorizing campaign practice and the future of election campaigns
Get PDFModern election campaigning is often described as “data-driven,” but there are signs that data may not always inform decision-making. This article focuses on the decision-making component of campaigns and provides a new theorisation of this previously neglected component of data-driven campaigns. Arguing that decision-making is affected by a range of contextual, agential and organizational factors, I call for qualitative analysis of campaign practice that is needed to understand how decisions are made. Applying this theoretical account, I consider the likely future of election campaigning, reflecting particularly on the potential for automation. Specifically, I identify where automation is likely to emerge and conclude that rote-like decision-making is unlikely to entirely define the practices of future campaigns. Collectively this article offers new theoretical insight into an overlooked aspect of modern campaigning - decision-making - and provides offers a foundation for more accurate predictions about the future of campaigns
Beyond species means – the intraspecific contribution to global wood density variation
Get PDFWood density is central for estimating vegetation carbon storage and a plant functional trait of great ecological and evolutionary importance. However, the global extent of wood density variation is unclear, especially at the intraspecific level. We assembled the most comprehensive wood density collection to date, including 109 626 records from 16 829 plant species across woody life forms and biomes (GWDD v.2, available here: doi: 10.5281/zenodo.16919509). Using the GWDD v.2, we explored the sources of wood density variation within individuals, within species and across environmental gradients. Intraspecific variation accounted for c. 15% of overall wood density variation (SD = 0.068 g cm−3). Variance was 50% smaller in sapwood than heartwood, and 30% smaller in branchwood than trunkwood. Individuals in extreme environments (dry, hot and acidic soils) had higher wood density than conspecifics elsewhere (+0.02 g cm−3, c. 4% of the mean). Intraspecific environmental effects strongly tracked interspecific patterns (r = 0.83) but were 70–80% smaller and varied considerably among taxa. Individual plant wood density was difficult to predict (root mean square error > 0.08 g cm−3; single-measurement R2 = 0.59). We recommend (1) systematic sampling of multiple individuals and tissues for local applications, and (2) expanded taxonomic coverage combined with integrative models for robust estimates across ecological scales
Enhancing safety of lithium-ion batteries in sustainable energy systems through intelligent minor short-circuits fault detection
Get PDFThe rapid growth of renewable energy integration and electric mobility has increased the demand for safe and reliable lithium-ion batteries, which are essential due to their high energy density, long lifespan, and efficiency. However, complex internal electrochemical reactions and external operational stress can induce minor short circuits (MSC) that are difficult to detect at early stages yet may escalate to thermal runaway, posing significant risks to large-scale energy storage systems. To address this challenge, this study proposes an unsupervised MSC fault diagnosis framework that integrates a hybrid feature extraction strategy with a deep support vector data description algorithm. The method employs two-dimensional correlation coefficients and two-dimensional wavelet transform to capture voltage consistency across cells and detect transient anomalies associated with fault development. These complementary features are fused into a multidimensional representation and processed by the deep model, which learns compact patterns of normal operating states and constructs a hypersphere for anomaly detection. The framework is validated using a laboratory module with six battery cells, demonstrating effective fault identification under varying operating conditions, fault severities, and battery chemistries, achieving a 94 % fault detection rate with a 3 % false alarm rate. Furthermore, the computational procedure relies on matrix-based feature construction and a lightweight feed-forward inference process, offering computational efficiency suitable for real-time deployment in battery management systems. Benefiting from its unsupervised and data-driven design, the framework exhibits strong generalizability under diverse conditions and provides a promising pathway for enhancing the safety and reliability of future energy storage applications
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇