A research team proposes a dynamic periodic synchronization strategy that adapts to different workloads, reducing synchronization costs. The strategy uses a quantitative analysis model and achieves the best tradeoff between consistency and synchronization cost.
Researchers have developed a novel noise correction method for crowdsourcing that considers both instance features and multiple noisy labels. The method, called LDSNC, outperforms existing competitors in reducing noise impact on integrated labels.
Researchers developed a compound metalens that enables distortion-free imaging across a wide field of view. The device demonstrated diffraction-limited performance and low barrel distortion, outperforming reference metalens.
Research highlights the importance of human and material capital in promoting sustainable tourism, while acknowledging unequal access to opportunities. The study recommends investing in public infrastructure and financial support for families to ensure equitable benefits.
Researchers developed a dual-layer optimization framework that integrates electricity and heat demand response, increasing revenue by 5.09% and consumer surplus by 2.46%. The strategy also reduced carbon trading costs by 5.23% and emissions by 2.54%, promoting environmental sustainability.
Graph convolutional networks improve collaborative patterns but exacerbate popularity bias, recommending more popular than lesser-known items. A new method corrects this by identifying and removing influential nodes' effects, improving recommendations for both popular and lesser-known items.
A recent study found that CIDEC/FSP27 enhances PVAT inflammation and macrophage infiltration, exacerbating AAA formation induced by high-fat diet and Ang II. FSP27 deficiency attenuates HFD-induced PVAT inflammation and prevents AAA formation.
Researchers report two cases of acquired immune thrombotic thrombocytopenic purpura (TTP) after administration of inactivated COVID-19 vaccine CoronaVac. TTP incidence increased in years following COVID-19 vaccination campaigns, suggesting a possible correlation between the vaccines and the condition.
Researchers propose a Graph-Segmenter using graph transformer and boundary-aware attention for simultaneous global and local view modeling. The approach enhances backbone capabilities by addressing issues in sliding window-based semantic segmentation.
A recent study by researchers from Zhejiang University and Nankai University estimates that AI systems could emit more than 102 million tons of CO₂ annually. The study analyzes energy consumption from 79 major AI systems between 2020 and 2024, highlighting the need for regulatory measures and standardized emissions caps.
A large body of research has been dedicated to automated issue classification for Issue Tracking Systems (ITSs). The team proposed an advanced approach called DeepLabel, which achieved better performance compared to existing approaches.
A recent study has identified the pivotal role of CD39 expression in modulating CAR-T cell function. Moderate CD39 expression levels displayed superior antitumor activity compared to those with high or low CD39 levels. The combination strategy of mdivi-1 and CD39 knockdown significantly enhanced the antitumor response.
Researchers evaluated four M-protease inhibitors against SARS-CoV-2 variants and other coronaviruses. LTV demonstrated superior inhibitory activity against M-protease mutants, while ETV showed limited impact on α-coronaviruses.
Delegable zk-SNARKs with proxies enable secure proof generation by sharing partial witnesses among multiple parties, achieving computational soundness and perfect zero knowledge.
Researchers discovered epigenetic silencing of BEND4 as a synthetic lethal marker for ATM inhibitors in pancreatic cancer. This approach targets the tumor suppressor gene, enhancing treatment efficacy. The study found significant association between BEND4 methylation and poor tumor differentiation.
Researchers successfully fabricate subwavelength nanostructures on Au nanofilm using laser direct-writing system, achieving minimum linewidth of 83.6 nm and repeatable linewidth of approximately 167.8 nm. The mechanism involves melting of the Au nanofilm due to locally excited surface plasmon polaritons.
A recent study using single-cell RNA sequencing has identified a 'starter' subpopulation of tumor cells responsible for initiating metastasis in neuroblastoma. The research also uncovered a 150-gene signature associated with poor prognosis, which may inform future therapeutic strategies.
A new research has proposed a Byzantine-Resilient and Verifiable Aggregation for Differentially Private Federated Learning, addressing the limitations of existing protocols. The approach achieves privacy-preservation, byzantine-resilience and verifiability through three primary methods: DPFA, DPLoss, and DPVeri.
A research team has proposed a distributed federated intrusion detection method using a blockchain-based architecture, which ensures secure training and discovery of new attack types. The proposed method outperforms existing methods in guaranteeing training security and discovering new attack types.
Researchers develop ReRAM-based PIM architecture ARCHER to process compressed recommendation systems on monolithic ReRAM chips. The approach enables large practical models on resource-constrained devices while outperforming existing solutions in terms of performance and energy savings.
A groundbreaking study reveals lncRNA Gm20257's regulatory role in pathological cardiac hypertrophy through its interaction with PGC-1α and the mitochondrial complex IV axis. Overexpression of Gm20257 reduces hypertrophic responses and improves cardiac function.
A new method, MLP-Mixer and Mixture of Expert model, predicts lithium-ion battery lifespan with high accuracy by handling varying contributions of different time steps. The proposed model outperforms existing methods on NASA and CALCE public datasets, providing reliable RUL predictions.
A collaborative team developed CShaperApp to segment and analyze Caenorhabditis elegans embryo morphologies, providing a user-friendly platform for researchers. The software offers modular functions for image segmentation, analysis, and result visualization, allowing users to adaptively retrain models or process external datasets.
Researchers found that chronic cold exposure reverses feeding-regulated LPL activity during browning in WAT, highlighting the hepatic insulin signaling pathway's role. This process involves hepatic secretion factors ANGPTL3 and ANGPTL8, offering new insights into energy expenditure and obesity prevention.
Bispecific ADCs leverage bispecific antibodies targeting multiple antigens, facilitating receptor clustering and internalization to release cytotoxic payloads within tumor cells. Several bispecific ADCs are in clinical development for various cancers, showing promising results.
A new virtual vibration test rig (VTR) simulates the fatigue life of dozer push arms with unprecedented accuracy, offering a cost-effective alternative to traditional physical testing. The VTR reduces testing time from hours to minutes while cutting costs, making it a game-changer for the construction industry.
X-ray detectors have achieved ultra-high sensitivity but still require improvement in detective quantum efficiency (DQE) for high-quality imaging. Researchers analyzed the influences of various factors on DQE and established requirements for optimal detector performance and circuit design.
Harnessing light's unique properties, photonic quantum computers exponentially accelerate computational tasks in various applications. Their practical uses extend to healthcare, AI, secure communication protocols, and precise molecular simulations vital for drug discovery.
Researchers used quantum mechanics/molecular mechanics and molecular dynamics to investigate the interaction between ATP and insulin-degrading enzyme in Alzheimer's disease. The study revealed crucial information for developing IDE inhibitors against ATP interactions, potentially slowing down AD progression.
Researchers found that multi-source data linkage enhances reliability of population-based cancer survival estimates in a metropolitan Chinese cancer registry. The study's findings have significant public health implications for optimizing cancer patient follow-up strategies in China and other developing countries.
A novel approach has improved the electrochemical performance of direct ammonia protonic ceramic fuel cells by introducing a CeO2-supported Ni and Ru catalyst layer. The study demonstrates enhanced performance and reduced degradation rates at various temperatures, paving the way for more efficient energy conversion systems.
Researchers developed predictive models using clinical characteristics to forecast treatment outcomes for small cell lung cancer patients. The models showed high accuracy in predicting response rates and disease control, providing a valuable decision-making aid for clinicians.
Research highlights significant growth in global cervical cancer research output, particularly in China, where hotspots focus on pathogenesis, epidemiology, HPV vaccination, and screening strategies. International cooperation and public health awareness are emphasized as key areas for improvement.
Global wildfires are on the rise, driven by climate change, contributing to carbon emissions and particulate matter. High-latitude regions face more intense wildfire events, highlighting the need for in-depth research to understand and mitigate their impacts.
Researchers identified distinct global patterns in human sewage microbiomes, with species richness increasing at higher latitudes. Economic factors, such as export values and social security, play a key role in shaping microbial diversity, with developed regions exhibiting more uniform communities.
Researchers have developed four advanced technologies to break down PFAS compounds, including hydrothermal alkali treatment, low-temperature mineralization, mechanochemical degradation, and adsorption. These methods demonstrate high efficiency in addressing the widespread environmental and health risks associated with PFAS.
A novel variant of tissue-like P systems with polarizations is proposed to solve the SAT problem in polynomial time, achieving theoretical feasibility for precise rule control in specific applications.
A recent study by Chenhao JIA and colleagues reveals a critical flaw in the DBST cipher, allowing attackers to deduce plaintext-ciphertext pairs without querying the encryption engine. The team proposes modifications to both linear and nonlinear parts of DBST to prevent this structure attack, resulting in improved diffusion performance.
Researchers proposed a new method, PC-AUC, to improve weakly-supervised learning frameworks for classification tasks. The method utilizes pairwise comparison data to maximize the area under the ROC curve (AUC), demonstrating its effectiveness on multiple datasets. Further leveraging unlabeled data could enhance performance in the future.
Researchers developed a Mo-doped NiFeOOH catalyst with significantly reduced overpotential, demonstrating superior catalytic activity. The catalyst's durability was also shown to be impressive, maintaining stable operation for up to 170 hours.
This study identifies Paenarthrobacter nitroguajacolicus as a potential biocontrol agent against cucumber corynespora leaf spot fungus, promoting cucumber growth and reducing disease severity. BJ-5 inhibited the germination of fungal spores and caused atrophy in budding tubes, with an inhibition efficacy rate of 63%.
A new research on serial-autoencoder-based personalized recommendation was published in Frontiers of Computer Science. The team proposes to retain the reconstructed auxiliary information in the decoding layer to further enhance the rating information, resulting in improved representation learning and scalability. This approach aims to ...
Researchers investigate quantum query complexity of basic matroid problems and present asymptotically optimal quantum algorithms for some of them. They apply the quantum adversary method to prove lower bounds on quantum algorithms, demonstrating that fundamental matroid problems can achieve quantum speedup.
Research explores impact of exogenous additives on heavy metal passivation and nitrogen retention in pig manure composting. Microbial additives show promise in reducing bioavailability of heavy metals, while also improving nitrogen retention.
Researchers develop multifunctional material for adjustable color emission, enhancing white LED performance and paving way for visual thermometers. The phosphors' stability, repeatability, and thermochromism make them suitable for high-temperature safety markings.
The study reveals significant spatial differences in diversity, structure, and function of microbial communities in the rhizosphere of Boehmeria nivea L. Volcanic soil plays a key role in forming microbial community diversity. Bacterial generalist species dominate, while fungi show greater ability to break down complex organic polymers.
This study breaks rotational symmetry in a deformed Reuleaux-triangle resonator to achieve exceptional points, enhancing high-chirality mode and nanoparticle detection up to 4000 nm. The simplified system demonstrates superior sensor sensitivity compared to non-deformed counterparts.
A new low-light enhancement algorithm balances performance with inference speed by brightening images first and then correcting degradation factors. The algorithm outperforms existing methods in noise suppression and color cast correction while maintaining a smaller model size.
The study found that China's net greenhouse gas emissions from agriculture increased significantly during the COVID-19 pandemic, with methane emissions accounting for over 65% of the total. Optimizing fertilizer use and manure management practices can help reduce nitrous oxide emissions.
The study analyzed historical trends in sugarcane and sugar beet yield, agricultural input, carbon footprint, and economic benefits in China. By adopting optimized crop models, GHG emission reduction potential can be achieved by 32% and total annual cost can be reduced by 24% by 2030.