Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks139 from public data
- Automatic Spatially-Aware Fashion Concept Discovery341
This paper proposes an automatic spatially-aware concept discovery approach using weakly labeled image-text data from shopping websites, and decomposes the visual-semantic embedding space into multiple concept-specific subspaces to facilitate structured browsing and attribute-feedback product retrieval.
- LLM Lies: Hallucinations are not Bugs, but Features as Adversarial Examples326
It is demonstrated that nonsensical prompts composed of random tokens can also elicit the LLMs to respond with hallucinations, and an automatic hallucination triggering method is formalized as thehallucination attack in an adversarial way.
- MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Use244
MetaTool is introduced, a benchmark designed to evaluate whether LLMs have tool usage awareness and can correctly choose tools, and recommends that tool developers choose an appropriate rewrite model for generating new descriptions based on the downstream LLM the tool will apply to.
- 225
- 133
- Predicting remaining useful life of rolling bearings based on deep feature representation and long short-term memory neural network129
A new remaining useful life prediction approach based on deep feature representation and long short-term memory neural network is proposed, and a new criterion, named support vector data normalized correlation coefficient, is proposed to automatically divide the whole bearing life as normal state and fast degradation state.
- 1000× Faster Camera and Machine Vision with Ordinary Devices116
A spiking neural network-based machine vision system that combines the speed of the machine and the mechanism of biological vision, achieving high-speed object detection and tracking 1,000x faster than human vision is developed.
- The iMaterialist Fashion Attribute Dataset93
This work contributes to the community a new dataset called iMaterialist Fashion Attribute (iFashion-Attribute), constructed from over one million fashion images with a label space that includes 8 groups of 228 fine-grained attributes in total, which is the first known million-scale multi-label and fine- grained image dataset.
- 85
- Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspective78
Guru, a curated RL reasoning corpus of 92K verifiable examples spanning six reasoning domains built through domain-specific reward design, deduplication, and filtering to ensure reliability and effectiveness for RL training, and two models that achieve state-of-the-art performance among open models RL-trained with publicly available data.
- 71
- Deep learning-based evaluation of the relationship between mandibular third molar and mandibular canal on CBCT67
An automated approach based on CNNs for detection and evaluation of the M3 and MC on CBCT has been established, which can be utilized to improve diagnostic efficiency and facilitate the precision diagnosis and treatment of M3.
- Ta-Coated Titanium Surface With Superior Bacteriostasis And Osseointegration64
In vivo biological studies showed that Ta modification significantly promoted the osseointegration of implants by stimulating the expression of bone-forming proteins, which may provide some insights into clinical applications of Ta-coated Ti implants, especially in possibly infected situations.
- 63
- On the Trustworthiness of Generative Foundation Models: Guideline, Assessment, and Perspective60
A holistic framework for advancing trustworthiness in GenAI is established, paving the way for safer and more responsible integration of GenFMs into critical applications and releasing the toolkit for dynamic evaluation.
- 42
- 41
- Deep Association: End-to-end Graph-Based Learning for Multiple Object Tracking with Conv-Graph Neural Network40
An efficient end-to-end model, Deep Association Network (DAN), to learn the graph-based training data, which are constructed by spatial-temporal interaction of objects, up to the state-of-the-art methods without extra-dataset on MOT16 and DukeMTMCT.
- 37
- 37
- 33
- Spectrum-Effect Relationship in Chinese Herbal Medicine: Current Status and Future Perspectives32
This review proves the validity and development potential of the SER and provides a reference for the development and application of quality evaluation methods for CHM.
- A Siamese network-based tracking framework for hyperspectral video31
A novel deep hyperspectral tracker based on Siamese network (SiamHT) is presented, designed to extract the spatial and spectral semantic features, respectively, and fused to estimate the state of a target.
- GRAPH-BASED REGULARIZATION FOR REGRESSION PROBLEMS WITH HIGHLY-CORRELATED DESIGNS26
This work considers a high-dimensional regression setting in which a graph governs both correlations among the covariates and the similarity among regression coefficients, and proposes graph-based regularization that yields mean-squared error guarantees for a broad range of covariance graph structures.
- 25
- Deep Trajectory Post-Processing and Position Projection for Single & Multiple Camera Multiple Object Tracking24
Position Projection Network (PPN) is presented to convert trajectory position from local camera-coordinate to global world-coodrinate, which provides adequate and accurate temporal-spatial information for trajectory association.
- Immunotherapy in colorectal cancer: Statuses and strategies23
This review endeavors to offer a comprehensive overview of current strategies in CRC immunotherapy, critically analyze existing literature, underscore anticipated outcomes from ongoing clinical trials, and deliberate on the challenges and impediments encountered within the field of immunotherapy.
- Edaravone Maintains AQP4 Polarity Via OS/MMP9/β-DG Pathway in an Experimental Intracerebral Hemorrhage Mouse Model21
Edaravone can maintain AQP4 polarity expression by inhibiting the OS /MMP9/β-DG pathway after ICH, which is suggested to maintain the loss of AQP4 polarity, downregulated GFAP expression, and upregulated β-DG expression.
- 21
- 21
- 20
- 20
- BBA-NET: A Bi-Branch Attention Network For Crowd Counting19
A Bi-Branch Attention Network (BBA-NET) for crowd counting, which can integrate the pedestrian’s head and body information to enhance the feature expression ability of the density map.
- 18
- Exploratory study of an anti-PD-L1/TGF-β antibody, TQB2858, in patients with refractory or recurrent osteosarcoma and alveolar soft part sarcoma: a report from Chinese sarcoma study group (TQB2858-Ib-02)17
The combination of PD-L1 and TQB2858 did not significantly improve the ORR in patients with recurrent osteosarcoma, however, it improved immunogenic responses in ASPS, even after progression upon anti-PD-1/PD-L1 therapy, with an acceptable safety profile.
- Multi-Channel CNN Based Inner-Attention for Compound Sentence Relation Classification17
A novel Inatt-MCNN model is proposed to extract sentence features and classify relations by combining multi-channel CNN and Inner-attention mechanism and has superior performance and achieves the highest accuracy, especially on the CCCS dataset.
- 16
- Graph-Based Regularization for Regression Problems with Alignment and Highly Correlated Designs16
This work considers a high-dimensional regression setting in which a graph governs both correlations among the covariates and the similarity among regression coefficients – meaning there is alignment between the covariates and regression coefficients.
- The Global Antimicrobial Resistance Trends of Staphylococcus aureus and Influencing Factors15
This study globally monitored the large-scale longitudinal trend of AMR in S. aureus and examined the various human and environmental climate factors that influence the occurrence and spread of AMR in S. aureus to provide a theoretical framework for monitoring the longitudinal trend of ARGs.
- Comparison of bilateral differential characteristics of corneal biomechanics between keratoconus and normal eyes15
The bilateral asymmetry of corneal biomechanics was significantly increased in keratoconus compared with normal eyes, which may be helpful for the early detection of keratconus.
- Enhancing and Dissecting Crowd Counting by Synthetic Data15
A simulated crowd counting dataset CrowdX, which has a large scale, accurate labeling, parameterized realization, and high fidelity, is proposed, which shows significant performance improvements of the proposed streamlined and efficient benchmark network ESA-Net.
- Distinguishment of different varieties of rhubarb based on UPLC fingerprints and chemometrics14
The developed UPLC fingerprints offer a simple, reliable and specific approach for distinguishing different varieties of rhubarb from three varieties of rhubarb.
- TIDE: Test-Time Few-Shot Object Detection14
This work formalizes a novel FSOD task, referred to as test-time few-shot detection (TIDE), where the model is un-tuned in the configuration procedure, and introduces an asymmetric architecture for learning a support-instance-guided dynamic category classifier.
- Masked Graph Neural Networks for Unsupervised Anomaly Detection in Multivariate Time Series14
According to the experimental findings, MGUAD outperforms state-of-the-art anomaly detection methods and can learn the structure of the unobserved causality among sensors to detect anomalies.
- High-Level Task-Driven Single Image Deraining: Segmentation in Rainy Days14
A new deraining network that contains a semantic refinement residual network (SRRN) and a novel two-stage segmentation aware joint training method that outperforms state-of-the-art methods and achieves visually better reconstruction results and sufficiently good performance on semantic segmentation task.
- EnclaveFuzz: Finding Vulnerabilities in SGX Applications13
EnclaveFuzz is proposed, a multi-dimension structure-aware fuzzing framework that analyzes enclave sources to extract input structures and correlations, then generates fuzz harnesses that can produce valid inputs to pass sanity checks and design a new sanitizer to detect both SGX-specific vulnerabilities and typical memory corruption vulnerabilities.
- Deep Human-Interaction and Association by Graph-Based Learning for Multiple Object Tracking in the Wild13
This paper specially considers the interrelation cue between targets and proposes Human-Interaction Model (HIM) to extract interaction features between the tracked target and its surrounding and proposes an efficient end-to-end model, Deep Association Network (DAN), to optimize the association with graph-based learning mechanism.
- 13
- A New Method to Detect Buffalo Mastitis Using Udder Ultrasonography Based on Deep Learning Network12
A 1/4 level mastitis detection model is established which provides a theoretical basis for mastitis detection in buffaloes mostly raised by small farmers lacking mastitis diagnostic conditions in developing countries.
- Carbopol dispersed PAA-modified UIO-66 with high colloidal stability as a combination nano-adjuvant boosts immune response and protection against pseudorabies virus in mice and pigs12
Much higher protection rates were achieved with the U@PAA-Car-adjuvated PRV vaccine in mice and pigs challenge than those observed from the commercial adjuvant groups.
- Application of Multiple-Optimization Filtering Algorithm in Remote Sensing Image Denoising12
A multiple-optimization bilateral filtering algorithm based on edge detection and differential evolution methods, which does not require any parameter input, is proposed and revealed that the MOBF algorithm outperforms traditional algorithms for all three evaluation metrics.
- 12
- 11
- Interlayer Electrons Polarization of Asymmetric Metal Nanoclusters/g‐C3N4 for Enhanced Microwave Therapy of Pneumonia11
Structurally modulated asymmetric doping of 2D materials with interfacial spatial effects can provide efficient microwave disinfection and pneumonia therapy.
- A Contrast-Enhanced CT-Based Deep Learning System for Preoperative Prediction of Colorectal Cancer Staging and RAS Mutation11
A deep learning system was created based on the foundation of contrast-enhanced CT portal-phase imaging to preoperatively predict the stage and RAS mutation status of colorectal cancer patients, which will help clinicians choose the best treatment option to increase colore CT patients’ chances of survival and quality of life.
- 10
- Advancements in cellular immunotherapy: overcoming resistance in lung and colorectal cancer9
This mini-review highlights the latest advancements in tumor immunotherapy, with a focus on cellular therapies, and addresses the resistance mechanisms that hinder their effectiveness in lung and colorectal cancers.
- I Think, Therefore I am: Benchmarking Awareness of Large Language Models Using AwareBench8
It is argued that awareness is an essential aspect of trustworthiness for LLMs to enhance their interaction with humans while ensuring ethical responses, and four key dimensions of awareness are identified: capability, mission, emotion, and perspective.
- 8
- An Adaptive Spatial-Temporal Local Feature Difference Method For Infrared Small-Moving Target Detection8
The experimental findings presented in this paper reveal that the suggested strategy outperforms current state-of-the-art approaches in the detection of IR tiny-moving object detection.
- 8
- 8
- Machine learning-based prediction model for patients with recurrent Staphylococcus aureus bacteremia7
The prediction model for recurrent SAB, developed by the team, aids physicians in timely diagnosis and treatment of patients and aids physicians in timely diagnosis and treatment of patients.
- FLASH Irradiation Regulates IFN-β induction by mtDNA via Cytochrome c Leakage7
The enhanced difference of cytochrome c leakage between cancer cells and normal cells post FLASH irradiation indicates a potential mechanism of FLASH effect by regulating the apoptotic and inflammatory pathway.
- A Novel Point Cloud Registration Method Based on ROPNet7
A new loss function based on the cross-entropy and applied to the ROPNet point cloud registration model is designed and improved by adding the channel attention mechanism to make the network focus on both global and local important information, thus improving the registration performance and reducing the point cloudRegistration error.
- Improving Multi-agent Reinforcement Learning with Imperfect Human Knowledge7
This paper proposes a novel approach to utilize imperfect human knowledge to improve the performance of multi-agent reinforcement learning by leveraging logic rules, which can be seen as a popular form of human knowledge, as part of the action space in reinforcement learning.
- 6
- 6
- 5
- Path Planning Method and Control of Mobile Robot with Uncertain Dynamics Based on Improved Artificial Potential Field and Its Application in Health Monitoring5
An attractive potential field rotation method to overcome the limitation that traditional artificial potential fields tend to fall into local minima and define a new class of attractive potential fields to address the goals non-reachable with obstacles nearby (GNRON) and collisions caused by excessive attractive force at long distances from the target point.
- 5
- 5
- Central Similarity Hashing via Hadamard matrix.5
A new hash center network (HCN) that learns hashing functions by optimizing the central similarity w.r.t.\ these hash centers is devised and can be applied for both image and video hashing.
- 5
- DALDet: Depth-Aware Learning Based Object Detection for Autonomous Driving4
This work designs an efficient one-stage detection framework and seamlessly integrate depth cues into convolutional neural network by introducing depth-aware convolution and depth-aware average pooling, which effectively improve the detector's ability to perceive 3D space.
- 4
- 4
- Robust object tracking with scene-adaptive scheme in occlusion4
A scene adaptive tracking algorithm in occlusion is presented that achieves the average overlap precision (OP) of 72.2%, which improves the performance by 7.6% compared to the DSST.
- A Lightweight Feature Enhancement Model for UAV Detection in Real-World Scenarios3
This study proposes a lightweight feature enhancement model (LFEM) based on YOLOv5, introducing the Ghost module to improve UAV detection with fewer floating-point operations (FLOPs) and incorporating the SIMAM module to enhance feature representation, particularly for real-world scenarios.
- PiGW: A Plug-in Generative Watermarking Framework3
Plug-in Generative Watermarking (PiGW) is proposed as a general framework for integrating watermarks into generative images and embeds watermark information into the initial noise using a learnable watermark embedding network and an adaptive frequency spectrum mask.
- 3
- Longitudinal Morphometric Changes in the Corticospinal Tract Shape After Hemorrhagic Stroke3
A significant positive association between the ipsilesional-CST deformation and hematoma volume was observed and a significant temporal difference at the first time point suggests CST restoration over time.
- 3
- Pillar in Pillar: Multi-Scale and Dynamic Feature Extraction for 3D Object Detection in Point Clouds3
A multi-scale voxelization method and a decomposable dynamic convolutional layer that considers the shared and dynamic components when applying Convolutional filters at different positions of feature maps to handle the variation of density in point cloud data.
- 3
- VulShield: Protecting Vulnerable Code Before Deploying Patches2
VulShield is an automated temporary protection system that is capable of protecting both vulnerable kernels, and user-space programs running on them, and does not rely on recent system features like eBPF and Linux security modules.
- Digital Rebirth of Dongba Pattern: An Improved Active Contour Model for Pattern Contour Extraction2
The results indicate that the proposed improved active contour model has advantages in contour extraction visual quality and evaluation indicators compared to existing methods, demonstrating the effectiveness of the proposed method.
- Distilling Desired Comments for Enhanced Code Review with Large Language Models2
A dataset distillation method, Desiview, which can automatically construct a distilled dataset by identifying DRCs from a code review dataset and validation results indicate that Desiview4FA slightly outperforms Desiview4FT, while both models have significantly improved against the base models in terms of generating DRCs.
- Multi-view dual-channel graph convolutional networks with multi-task learning2
A novel approach is proposed to address the problem of insufficient information consideration in network embedding, which is termed multi-task-oriented adaptive dual-channel graph convolutional network (TAD-GCN), which contains dual- channel GCN which can extract the specific and shared embeddings from multiple views simultaneously.
- 2
- 2
- 2
- 2
- 2
- Particle Image Velocimetry Algorithm Based on Spike Camera Adaptive Integration1
Experimental results demonstrate that the proposed particle image velocimetry algorithm based on the adaptive integration of the spike camera effectively addresses overexposure challenges caused by high illumination of the liquid–gas interface in flow field measurements.
- Policies and Practices Regarding Preparative Fasting for Contrast-Enhanced Computed Tomography: A Nationwide Survey1
Preparative fasting policies varied among hospitals in terms of the fasting content and duration, and the latest guideline regarding no fasting before CECT has not been fully adopted.
- 1
- 1
- WUDA: Unsupervised Domain Adaptation Based on Weak Source Domain Labels1
This paper proposes two intuitive frameworks for unsupervised domain adaptation based on weak source domain labels (WUDA) and conducts comparative experiments to analyze the impact of domain shift changes on the two frameworks.
- 1
- Pharmacist-led iodinated contrast media infusion risk assessment service1
A pharmacist-led multidisciplinary team can provide advance warning and effectively limit the risks of adverse reactions caused by iodinated contrast media to a preventable and controllable level and is encouraged to implement in other areas of China.
- Relative vectors clustering and temporal constraint based generalized Hough transform for high frame rate and ultra-low delay arbitrary shape detection1
This paper proposes clustering relative-vectors-based parallelization and temporal constraint for generalized Hough transform (GHT) algorithm compression to achieve the ultra-low delay process system, implemented on FPGA.
- 1
- 1
- [Applications of Artificial Intelligence in Musculoskeletal System Imaging].1
The application of AI in musculoskeletal imaging is summarized and analyzed to improve the efficiency and accuracy of imaging diagnosis and reduce medical cost.
- 1
- 1
- 1
- Some Properties of D -Concurrence1
A new lower bound for the concurrence is given, which is related to D-concurrence, and for the Werner states, it is compared with the other known bound.
- A Hybrid Mesenchymal-Stem-Cell-Derived Decellularized Matrix Scaffold Supports Bone Repair and Vascular Perfusion in Steroid-Associated Osteonecrosis–
It is suggested that mesenchymal-stem-cell-derived CDM functions as a bioactive matrix component that helps rebalance the local inflammatory niche and supports bone repair with improved vascularization-related outcomes in SAON.
- The Comparison of Diagnostic Performance Between Next-Generation Sequencing of Blood and Tissues for Primary Spinal Infections–
The diagnostic performance of blood mNGS is worse than tissue mNGS in primary spinal infections and the specificity of blood mNGS was the lowest among the 4 pathogen identification techniques.
- –
- –
- Application of deep learning in analysing morphological parameters of cervical computed tomography scans–
A series of DL-based segmentation algorithms to automatically measure five key parameters for the evaluation of cervical spondylotic myelopathy can effectively approximate the manual measurements of cervical CT morphological parameters in humans, offing physicians an accurate and efficient auxiliary diagnostic tool.
- –
- –
- Apatinib combined with ifosfamide and etoposide versus ifosfamide and etoposide in relapsed or refractory osteosarcoma (OAIE/PKUPH-sarcoma 11): A multicenter, randomized controlled trial.–
Apatinib plus IE demonstrated a significant improvement in PFS in patients with advanced osteosarcomas, with an acceptable safety, as well as numerically higher disease control rates in the apatinib plus IE group.
- –
- –
- –
- DINA Model with Entropy Penalization–
This paper proposes an entropy-penalized likelihood method to eliminate the empty attribute profiles, and the relation between attribute profiles and the parameter space of item parameters is discussed, and two modified expectation–maximization algorithms are designed to estimate the model parameters.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- MSFormer: A Skeleton-multiview Fusion Method For Tooth Instance Segmentation–
2D-3D joint perception is advocated and MSFormer, a novel method for tooth segmentation that combines two lightweight modules into existing multiview-based models, achieves state-of-the-art performance, requiring only 100 training meshes.
- –
- –
- Single image super-resolution using deep hierarchical attention network–
A compact and accurate super-resolution algorithm using the attention-augmented convolutional neural network, which can exploit and weight hierarchical features at multiple scales and levels to improve learning capability is presented.
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.