6974 research outputs found
Sort by
3C:Confidence-guided clustering and contrastive learning for unsupervised person re-identification
No full textUnsupervised person re-identification (Re-ID) aims to learn a feature network with cross-camera retrieval capability in unlabelled datasets. Although pseudo-label based methods have achieved great progress in Re-ID, their effectiveness in complex scenarios is hindered by three unresolved challenges: (i) Noisy pseudo-labels due to disturbed clustering. (ii) Camera-biased clusters that lack diversity. (iii) Unreliable hard samples that amplify noise in contrastive learning. To address these issues, a Confidence-guided Clustering and Contrastive learning (3C) framework is proposed in this paper. The 3C framework presents three confidence degrees: (i) in the clustering stage, the confidence of the discrepancy between samples and clusters is proposed to implement a harmonic discrepancy clustering algorithm (HDC); (ii) in the forward-propagation training stage, the confidence of the camera diversity of a cluster is evaluated using a novel camera information entropy (CIE), and clusters with high CIE will play the leading role in model training; and (iii) in the back-propagation training stage, the confidence of the hard sample in each cluster is designed and further used in a confidence integrated harmonic discrepancy (CHD), to select the informative sample for updating the memory for contrastive learning. Extensive experiments on three popular Re-ID benchmarks demonstrated the superiority of the proposed framework. In particular, the 3C framework achieved state-of-the-art results: 86.7%/94.7%, 45.3%/73.1% and 47.1%/90.6% in term of mAP/Rank-1 accuracy on Market-1501, MSMT17 and VeRi-776, respectively.<br/
Quantum Time Travel Revisited:Noncommutative Möbius Transformations and Time Loops
No full textWe extend the theory of quantum time loops introduced by Greenberger and Svozil [1] from the scalar situation (where paths have just an associated complex amplitude) to the general situation where the time traveling system has multidimensional underlying Hilbert space. The main mathematical tool that emerges is the noncommutative Möbius Transformation and this affords a formalism similar to the modular structure well known to feedback control problems. The self-consistency issues that plague other approaches do not arise here, as we do not consider completely closed time loops. We argue that a sum-over-all-paths approach may be carried out in the scalar case but quickly becomes unwieldy in the general case. It is natural to replace the beam splitters of [1] with more general components having their own quantum structure, in which case the theory starts to resemble the quantum feedback network theory for open quantum optical models and indeed we exploit this to look at more realistic physical models of time loops. We analyze some Grandfather paradoxes in the new setting
Metal(oid)s in Ulva:should we be worried?
No full textUlva spp. are promising food resources owing to their nutritional richness and beneficial properties. However, it accumulates potentially toxic trace elements, raising health safety concerns and proving useful for biomonitoring studies. In response to this concern, this review, conducted in collaboration with the EU-COST Action CA 20106 network, critically analysed 176 peer-reviewed papers to evaluate metal(oid) accumulation in Ulva. This study revealed substantial variability in the essential and non-essential element content due to environmental conditions, geographic regions, morphological forms, and analytical methods used in both wild and cultivated Ulva. The analysis was based on gross morphology (tube or foliose) rather than species-level identification. The identification of toxic forms, such as methylmercury and inorganic arsenic, remains limited, highlighting the need for element speciation to more accurate assess safety. Based on these findings, the review identified and outlined key areas requiring attention to ensure the safe and effective use of Ulva. Standardised analytical protocols are needed to improve consistency and comparability across studies and to enable accurate detection of toxic element forms. Improved taxonomic resolution, using molecular tools, is essential for distinguishing species-specific accumulation patterns. Expanding research into understudied geographic regions will help capture global variability in environmental influences on trace element uptake. Finally, standardised cultivation parameters are crucial to control elemental composition in farmed Ulva and to ensure its suitability for human consumption and commercial applications.</p
Insect protein to support human skeletal muscle anabolism:A systematic review of randomised controlled trials
No full textCONTEXT: A global shift toward sustainable food sources is emerging due to the immense environmental pressure from the production of animal foods. Insects present a novel source of sustainable dietary protein, due to their high protein content and favourable amino acid profile.OBJECTIVES: The aim of this systematic review was to establish the effects of insect protein compared with animal protein on skeletal muscle anabolism and adaptation. This review also explores the usefulness of insects for supporting the protein needs of population groups with high protein requirements.DATA SOURCES: Database searches were performed using the search terms "edible insects" and "insect protein," plus the key words "human health," "exercise," "anabolic response," "muscle protein synthesis," "skeletal muscle," "muscle adaptation," "lean mass," and "bioavailability." Studies had to be randomised controlled trials conducted with adult human participants (aged >18 years) that measured protein bioavailability, anabolic response, or skeletal muscle adaptation, with direct comparison between insect and animal protein.DATA EXTRACTION: Four studies (n = 100 participants) were included in the review. Of the two studies that assessed only postprandial blood amino acid concentration, one reported higher aminoacidemia from cricket compared with beef protein ingestion, and the other reported higher aminoacidemia from whey compared with lesser mealworm protein ingestion. Two studies also directly assessed the postprandial skeletal muscle anabolic response after exercise. Both reported lower peak plasma amino acid concentration from cricket or lesser mealworm protein compared with whey or milk protein, but there was no difference in skeletal muscle anabolism between the insect and animal protein sources.CONCLUSION: Insects are a viable protein source that can likely support skeletal muscle anabolism to the same extent as conventional animal protein but with a considerably lower environmental impact. Insects could be an effective protein source to facilitate skeletal muscle during challenging life circumstances or for those with physically demanding occupations.</p
Welsh Cultural Identity and the Eighteenth-Century Evangelical Revival
No full textSignificant developments in the eighteenth and nineteenth centuries led to a new definition in public discourse of what it meant to be Welsh for a substantial proportion of the population. This work explores the nature of the cultural identity which emerged through the influence of the evangelical revival, making use of the works of William Williams, Pantycelyn, to consider how that identity was fashioned and what its lasting impact was
Understanding contraception-use intentions among women of reproductive age not currently using contraceptives in sub-Saharan Africa:Key insights from Demographic and Health Surveys
Get PDFBackgroundThis study assesses the prevalence of contraception-use intentions and evaluates the associated factors among non-users in sub-Saharan Africa (SSA).MethodsData from 2014–2023 Demographic and Health Surveys of 30 countries in SSA consisting of 332 986 women aged 15–49 y not already using contraception were used.ResultsThe overall prevalence was 41.18% (95% CI 41.01 to 41.34%). Zimbabwe had the highest prevalence (72.34%; 95% CI 71.11 to 73.57%), whereas Ethiopia had the lowest (15.96%; 95% CI 15.40 to 16.51%). Women aged 25–49 y had lower odds of intending to use contraception compared with those aged 15–19 y, and this was striking among those aged 45–49 y (adjusted OR [AOR]=0.06, 95% CI 0.06 to 0.07). Those with a higher level of education displayed a greater likelihood of intending to use contraception (AOR=1.93, 95% CI 1.82 to 2.05) compared with those with no education. The odds increased with the number of children born, particularly for those with ≥4 children (AOR=1.59, 95% CI 1.52 to 1.67) compared with those with no children.ConclusionsPromoting the use of contraception requires tailored, multi-pronged interventions that account for the diverse sociodemographic, fertility and informational needs of women in this population
Unsupervised multimodal thick cloud removal for optical remote sensing images via adversarial learning
No full textCloud contamination is a common degradation in optical remote sensing images, adversely affecting the application of such images. Deep-learning-based cloud removal algorithms with auxiliary information have received increasing attention in recent years. Most of these methods rely on georeferenced, cloud-free optical images from other periods as references. However, the inherent gap between the reference and the target images often leads to inaccurate reconstruction. Unsupervised methods have also been proposed, mitigating the gap issue by eliminating the need for reference images. Yet, they typically and solely rely on reconstruction loss during training, often resulting in unnatural outcomes. To tackle these limitations, we propose ALM-CR (Adversarial Learning–based Multimodal Cloud Removal), an unsupervised two-stage framework that leverages synthetic aperture radar (SAR) as auxiliary input. The first stage performs SAR-to-optical translation for structural and approximate spectral recovery, followed by SAR-optical fusion to restore fine-grained spectral details. The proposed adversarial learning strategy removes the need for temporal reference images, enabling precise reconstruction of cloud-covered images while preventing overfitting. Experimental results demonstrate that our method surpasses existing unsupervised methods on both reference and no-reference metrics, and reconstructs spectral information more consistently than supervised methods.</p
QuTiP 5:The Quantum Toolbox in Python
Get PDFQuTiP, the Quantum Toolbox in Python (Johansson et al., 2012, Johansson et al., 2013), has been at the forefront of open-source quantum software for the past 13 years. It is used as a research, teaching, and industrial tool, and has been downloaded millions of times by users around the world. Here we introduce the latest developments in QuTiP v5, which are set to have a large impact on the future of QuTiP and enable it to be a modern, continuously developed and popular tool for another decade and more. We summarize the code design and fundamental data layer changes as well as efficiency improvements, new solvers, applications to quantum circuits with QuTiP-QIP, and new quantum control tools with QuTiP-QOC. Additional flexibility in the data layer underlying all “quantum objects” in QuTiP allows us to harness the power of state-of-the-art data formats and packages like JAX, CuPy, and more. We explain these new features with a series of both well-known and new examples. The code for these examples is available in a static form on GitHub (https://github.com/qutip/qutip-paper-v5-examples) and as continuously updated and documented notebooks in the qutip-tutorials package (https://github.com/qutip/qutip-tutorials).</p
Trust-enhanced POI recommendation algorithm using expectation-maximization
Get PDFPoint-of-interest (POI) recommendation systems have become increasingly important as travelers rely on mobile technologies and location-based social networks to discover new places. However, existing approaches often struggle with static user preferences, inadequate trust modeling, and extreme data sparsity. This paper introduces ExMax, a dynamic trust-enhanced recommendation framework leveraging Expectation-Maximization theory to address these limitations. ExMax employs a novel check-in matrix representation that adapts to evolving user interests, incorporates friendship network information to enhance recommendation quality, and integrates sentiment analysis to capture nuanced satisfaction signals beyond ratings. The framework’s iterative probabilistic model discovers latent features within sparse data, enabling meaningful recommendations even with limited interaction history. The algorithm exhibits time complexity for offline learning, where T represents EM iterations (typically 20–30), |R| denotes observed ratings, F indicates features, and K represents gradient steps. While sparse matrix operations and parallelization potential provide some mitigation, the iterative nature poses scalability challenges for platforms with hundreds of millions of users. Experimental evaluation on Yelp, Gowalla, and Brightkite datasets demonstrates that ExMax performs favorably compared to existing approaches across various metrics. The results suggest that integrating dynamic preference modeling, social trust signals, and contextual information offers a promising direction for location-based recommendation systems, particularly where recommendation quality justifies the computational cost. This work demonstrates how probabilistic modeling can effectively capture the dynamic and social nature of location discovery while acknowledging the inherent computational trade-offs of iterative optimization.</p
Effects of sensory and environmental labelling of plant-based products on consumer acceptance:Context, energy density and framing factors
Get PDFThere is growing pressure to replace animal-sourced proteins with plant-based proteins. Consumer studies suggest sensory properties and environment are the major factors impacting adoption of PBFs, but few studies have contrasted these factors. Knowing that health labels negatively impact sensory experience, we tested whether environmental labels had the same negative impact. Using an online survey, volunteers (N = 328) were randomly assigned to one of three label contexts: sensory (emphasizing taste and texture), environmental (highlighting sustainability and environmental impact), or control (no specific messaging), where they evaluated eight plant-based alternative foods. Each product was enhanced by either a positive or a negatively valanced framing statement, with half the foods higher, and half lower, in energy density (ED). Participants rated expected liking, wanting and likely recommendation, and estimated what they would pay for each food. For liking and recommending, there was no significant difference between environmental and sensory contexts (p = 0.94), but both were significantly higher than control (p = 0.0006), while for expected wanting only the sensory exceeded the control (p = 0.0014). The amount willing to pay was significantly higher in the environmental than sensory (p = 0.0005) or control (p < 0.0001) contexts, which did not differ significantly (p = 0.49). For all four measures, higher ED foods were rated significantly more positively than lower ED (p < 0.001), while the effect of environment on purchase price was magnified by higher ED foods (p < 0.001). Positive framing statements were rated significantly higher than negative framing for liking (p < 0.001), wanting (p < 0.001) and recommending (p = 0.022), but not for purchase (p = 0.30). When habitual diet (plant-based or not) was included in the exploratory analyses, it only altered acceptance of lower energy-dense products in the control context. Overall, these data suggest that the use of environmental descriptors may enhance consumer expectations and willingness to pay more to the same degree as sensory descriptors, providing various strategies for marketers and product developers to promote PBFs based on messages that best fit the brand identity and expand the PBFs narrative beyond health