Zhejiang University

An identification and classification framework for bridge cracks using transposed convolution-CBAM-Segformer with transfer learning

Bridge cracks seriously affect structural bearing capacity, durability and traffic safety, making accurate and efficient crack detection essential for bridge maintenance. To address the problems of feature degradation and upsampling information loss in existing segmentation model...

An analysis of spatial mismatch and coordination mechanisms in river-sea systems

China's port system within the Yangtze River Economic Belt has evolved from a traditional coastal port–hinterland structure into a more interactive coastal–inland port network. Coastal ports such as Shanghai port now operate as interconnected nodes, where coordination and investm...

Sub-Poissonian Electroluminescence from Resonant Inelastic Tunneling in One-Dimensional Gold Atomic Chains

Abstract Electrically driven single-photon sources are central to integrated quantum photonics, but Poissonian carrier injection usually transfers electrical shot noise into classical photon statistics. Here we propose to suppress this bottleneck at its origin by using a one-dime...

Biochar–Straw Interaction Drives Soil Carbon Accrual by Modulating Microbial Metabolic Pathways

Abstract Straw and biochar additions can enhance long-term carbon storage, yet how their interactions regulate carbon mineralization and allocation remains unclear. We conducted a 360-day microcosm incubation employing a reciprocal 13C-labeling design that paired 13C-labeled stra...

DDOR: Delta Debugging for Explainable Overrefusal Testing and Repair

While safety alignment and guardrails help large language models (LLMs) avoid harmful outputs, they can also induce overrefusal, i.e., unwarranted rejection of benign queries that merely appear risky. We present DDOR (Delta Debugging for OverRefusal), a fully automated and explai...

Spatial responses of ecosystem services to future climate change across China in the mid-21st century

Accurately identifying ecosystem-service (ES) responses to future climate change is essential for improving ecological conservation and restoration and for developing targeted adaptation strategies. However, coherent future response zones across multiple regulating services and t...

Don’t Use a Cannon to Kill a Fly: Lightweight Model Editing for LLMs to Correct Deprecated API Recommendations

Pre-trained or fine-tuned on large code corpora, Large Language Models (LLMs) have demonstrated strong performance in code completion tasks. However, their embedded knowledge is constrained by the timeliness of training data, which often includes code using deprecated APIs. Conse...

An Empirical Study of Speculative Decoding on Software Engineering Tasks

Large Language Models (LLMs) have become widely used for Software Engineering (SE) tasks, spanning from function-level code generation to complex repository-level workflows. However, the high latency of autoregressive inference remains a significant bottleneck, hindering their de...

Better Call Grep: Evaluating and Improving Grep-Like Lexical Retrieval for Repository-Level Code Completion

Repository-level code completion remains challenging for large language models (LLMs), as it requires reasoning over cross-file dependencies while under limited context windows. To address this challenge, prior work has adopted Retrieval-Augmented Generation (RAG) frameworks base...

Answer Is Cheap, Show Me the Evidence! Augmenting Automated Vulnerability Assessment with Evidence

Software Vulnerability (SV) assessment is a vital phase in SV management, which characterizes discovered SVs to locate hot spots and prioritize their remediation. To reduce the overhead and latency of manual assessment, prior works have explored automatically predicting assessmen...

SmartDecompiler-R1: Enhancing Faithful and Explainable Smart Contract Bytecode Decompilation with Reinforcement Learning

Understanding EVM bytecode is critical for smart contract security analysis. Existing decompilers typically rely on heuristic rules or leverage large language models (LLMs) to generate source code after bytecode analysis. However, heuristic-based approaches often produce pseudoco...

Multiple Benefits of Deploying Wastewater Phosphorus Recovery Strategies in China: A Spatially Explicit Scenario Assessment

Abstract Phosphorus (P) management faces dual pressures from finite mineral resources and waterborne pollution. Recovering P from wastewater offers a strategic pathway to alleviate both constraints simultaneously, yet systematic evaluations remain scarce, particularly with respec...

Nonthermalized Multiple Localized States Enable Efficient and Stable White-Light-Emitting Quantum Dots

Abstract Broadband emission from single quantum dot (QD) inherently relies on localized states that are conventionally regarded as nonradiative trap manifolds and suffer from low synthetic reproducibility. Here we show that photogenerated holes in a AgInxGa1–xS2/ZnS core/shell QD...

On the Evaluation of Large Language Models in Unit Test Evolution (Experience Paper)

Large language models (LLMs) have recently shown promising potential in automating unit test evolution for evolving software systems. However, the effectiveness of LLMs in unit test evolution remains insufficiently understood, particularly with respect to prompt design choices, i...

Single-cell profiling and genetic regulation of alternative polyadenylation in the human brain

Abstract Alternative polyadenylation (APA) has a central role in regulation of the fate of mRNA and exhibits high variability in the brain and in neurons. However, brain APA regulation has not been mapped at cell type resolution, and the genetic control of APA in brain cell types...

The exact maximum growth factor for complete pivoting on real 5 × 5 matrices

AbstractLet ρ₅ᴿ be the largest element growth possible in exact Gaussian elimination with complete pivoting on real 5 × 5 matrices, with every legal choice at a pivot tie allowed. We prove ρ₅ᴿ = α = 4.132517078632472854223346853277…, where α is the algebraic lower bound identifie...

Neddylation E2 enzyme UBE2M inhibits inflammatory lung tumor microenvironment by blocking the cGAS-STING signal upon KrasG12D activation.

activation in the lung.

Large-scale genetic profiling of hereditary ataxias in China: implications for a stepwise molecular diagnostic strategy

BACKGROUND: Hereditary ataxias are clinically and genetically heterogeneous, which complicates clinical management, genetic counseling, and scientific research. Efficient molecular diagnosis is essential but requires molecular epidemiological considerations due to the variable ge...

Prevalence and geographic variation of enteric pathogens among asymptomatic individuals in China: a nationwide, cross-sectional, multiplex molecular survey

The increasing use of multiplex molecular assays has improved the detection of enteric pathogens, but baseline carriage in asymptomatic populations remains poorly characterized, complicating the interpretation of positive test results. We aimed to investigate the prevalence and g...

Microscopic experimental investigation into the effects of silt on gas hydrate formation in sandy marine sediments

Marine gas hydrate reservoirs are rich in fine-grained silt, which substantially affects gas hydrate occurrence patterns and sediment pore characteristics. The presence of silt is a key factor that controls gas production efficiency and sand production behavior during hydrate exp...