Yanfeng Wang
YOU?
Author Swipe
View article: VocalBench-DF: A Benchmark for Evaluating Speech LLM Robustness to Disfluency
VocalBench-DF: A Benchmark for Evaluating Speech LLM Robustness to Disfluency Open
While Speech Large Language Models (Speech-LLMs) show strong performance in many applications, their robustness is critically under-tested, especially to speech disfluency. Existing evaluations often rely on idealized inputs, overlooking c…
View article: DICE: Structured Reasoning in LLMs through SLM-Guided Chain-of-Thought Correction
DICE: Structured Reasoning in LLMs through SLM-Guided Chain-of-Thought Correction Open
When performing reasoning tasks with user-specific requirements, such as strict output formats, large language models (LLMs) often prioritize reasoning over adherence to detailed instructions. Fine-tuning LLMs on supervised datasets to add…
View article: HeteroRAG: A Heterogeneous Retrieval-Augmented Generation Framework for Medical Vision Language Tasks
HeteroRAG: A Heterogeneous Retrieval-Augmented Generation Framework for Medical Vision Language Tasks Open
Medical large vision-language Models (Med-LVLMs) have shown promise in clinical applications but suffer from factual inaccuracies and unreliable outputs, posing risks in real-world diagnostics. While retrieval-augmented generation has emer…
View article: Improving optimal prompt learning through multilayer fusion and latent dirichlet allocation
Improving optimal prompt learning through multilayer fusion and latent dirichlet allocation Open
Recent advances in few-shot learning have demonstrated the potential of prompt-based techniques with pre-trained models, eliminating the need for extensive fine-tuning. However, challenges such as obtaining optimal prompts and addressing d…
View article: ConText: Driving In-context Learning for Text Removal and Segmentation
ConText: Driving In-context Learning for Text Removal and Segmentation Open
This paper presents the first study on adapting the visual in-context learning (V-ICL) paradigm to optical character recognition tasks, specifically focusing on text removal and segmentation. Most existing V-ICL generalists employ a reason…
View article: Impact of gene-environment interactions on atrial fibrillation and cardiac structure
Impact of gene-environment interactions on atrial fibrillation and cardiac structure Open
View article: Optimizing Reinforcement Learning with Limited HRI Demonstrations: A Task-Oriented Weight Update Method with Analysis of Multi-Head and Layer Feature Combinations
Optimizing Reinforcement Learning with Limited HRI Demonstrations: A Task-Oriented Weight Update Method with Analysis of Multi-Head and Layer Feature Combinations Open
To address the challenge of training reinforcement learning (RL) networks with limited data in Human-Robot Interaction (HRI), we introduce a novel task-oriented update method that combines meta-inverse reinforcement learning (Meta-IRL) and…
View article: Image encryption scheme based on thorp shuffle and pseudo dequeue
Image encryption scheme based on thorp shuffle and pseudo dequeue Open
View article: Product competitor identification analysis based on sentiment analysis
Product competitor identification analysis based on sentiment analysis Open
View article: Enhancing prediction and stratifying risk: machine learning and bayesian-learning models for catheter-related thrombosis in chemotherapy patients
Enhancing prediction and stratifying risk: machine learning and bayesian-learning models for catheter-related thrombosis in chemotherapy patients Open
While ML models demonstrated high predictive performance, their clinical applicability was limited due to complexity. The Bayesian-learning-based risk stratification model provided a simplified yet robust alternative, effectively predictin…
View article: Ethical-Lens: Curbing malicious usages of open-source text-to-image models
Ethical-Lens: Curbing malicious usages of open-source text-to-image models Open
The burgeoning landscape of text-to-image models, exemplified by innovations such as Midjourney and DALL·E 3, has revolutionized content creation across diverse sectors. However, these advances bring forth critical ethical concerns, partic…
View article: Association between life's essential 8 and Parkinson's disease: a case–control study
Association between life's essential 8 and Parkinson's disease: a case–control study Open
Objectives Life's essential 8 (LE8) is an emerging approach for accessing and quantifying cardiovascular health (CVH), but the effect on Parkinson's disease (PD) is still unclear. This study aimed to elucidate the association between LE8 m…
View article: Joint effect of modifiable risk factors on Parkinson’s disease: a large-scale longitudinal study
Joint effect of modifiable risk factors on Parkinson’s disease: a large-scale longitudinal study Open
Introduction Previous researches have often underestimated the diversity and combined effects of risk factors for Parkinson’s disease (PD). This study aimed to identify how multiple modifiable risk factors collectively impact PD. Methods T…
View article: VARFVV: View-Adaptive Real-Time Interactive Free-View Video Streaming with Edge Computing
VARFVV: View-Adaptive Real-Time Interactive Free-View Video Streaming with Edge Computing Open
Free-view video (FVV) allows users to explore immersive video content from multiple views. However, delivering FVV poses significant challenges due to the uncertainty in view switching, combined with the substantial bandwidth and computati…
View article: Towards Omni-RAG: Comprehensive Retrieval-Augmented Generation for Large Language Models in Medical Applications
Towards Omni-RAG: Comprehensive Retrieval-Augmented Generation for Large Language Models in Medical Applications Open
Large language models hold promise for addressing medical challenges, such as medical diagnosis reasoning, research knowledge acquisition, clinical decision-making, and consumer health inquiry support. However, they often generate hallucin…
View article: Cost: Contrastive One-Stage Transformer for Vision-Language Small Object Tracking
Cost: Contrastive One-Stage Transformer for Vision-Language Small Object Tracking Open
View article: DICE: Structured Reasoning in LLMs through SLM-Guided Chain-of-Thought Correction
DICE: Structured Reasoning in LLMs through SLM-Guided Chain-of-Thought Correction Open
View article: MobileA3gent: Training Mobile GUI Agents Using Decentralized Self-Sourced Data from Diverse Users
MobileA3gent: Training Mobile GUI Agents Using Decentralized Self-Sourced Data from Diverse Users Open
View article: A real-world pharmacovigilance analysis of potential ototoxicity associated with sacubitril/valsartan based on FDA Adverse Event Reporting System (FAERS)
A real-world pharmacovigilance analysis of potential ototoxicity associated with sacubitril/valsartan based on FDA Adverse Event Reporting System (FAERS) Open
Sacubitril/valsartan, a first-in-class angiotensin receptor neprilysin inhibitor, is widely used to treat heart failure. Despite its efficacy, sacubitril/valsartan inevitably causes adverse events such as hypotension, renal dysfunction, hy…
View article: Acceleration Harmonic Estimation and Suppression for Hydraulic Load Simulator Based on Artificial Bee Colony with Chaotic Search Strategy
Acceleration Harmonic Estimation and Suppression for Hydraulic Load Simulator Based on Artificial Bee Colony with Chaotic Search Strategy Open
This paper presents a new hybrid algorithm based on artificial bee colony (ABC) and chaotic search strategy (CSS), known as ABC-CSS to solve the harmonic estimation problems in the case of time varying acceleration response signal. The ABC…
View article: Performance Characteristics of Newly Developed Real-Time Wave Measurement Buoy Using the Variometric Approach
Performance Characteristics of Newly Developed Real-Time Wave Measurement Buoy Using the Variometric Approach Open
Accurate measurement of ocean wave parameters is critical for applications including ocean modeling, coastal engineering, and disaster management. This article introduces a novel global navigation satellite system (GNSS) drifting buoy for …
View article: Molecular probes for tracking lipid droplet membrane dynamics
Molecular probes for tracking lipid droplet membrane dynamics Open
View article: HPC: Hierarchical Progressive Coding Framework for Volumetric Video
HPC: Hierarchical Progressive Coding Framework for Volumetric Video Open
Volumetric video based on Neural Radiance Field (NeRF) holds vast potential\nfor various 3D applications, but its substantial data volume poses significant\nchallenges for compression and transmission. Current NeRF compression lacks the\nf…
View article: HLX07 alone or combined with serplulimab, cisplatin and 5‐fluorouracil for advanced esophageal squamous cell carcinoma: A phase 2 study
HLX07 alone or combined with serplulimab, cisplatin and 5‐fluorouracil for advanced esophageal squamous cell carcinoma: A phase 2 study Open
Background The combination of anti‐PD‐1 antibody serplulimab and chemotherapy is considered standard first‐line therapy for advanced esophageal squamous cell carcinoma (ESCC), but few later‐line treatments are available. Here we evaluated …
View article: Robust finite-time output feedback stabilisation for a class of uncertain planar systems with asymmetric output constraints
Robust finite-time output feedback stabilisation for a class of uncertain planar systems with asymmetric output constraints Open
View article: Screening and Characterization of Wild Sarcomyxa edulis Strains from Heilongjiang, China, for Strain Development
Screening and Characterization of Wild Sarcomyxa edulis Strains from Heilongjiang, China, for Strain Development Open
Sarcomyxa edulis is a characteristic low-temperature, edible mushroom in Northeast China. It has a delicious taste and rich nutritional and medicinal value. The artificial cultivation of S. edulis has been increasing in recent years. Howev…
View article: Breast Tumor Diagnosis Based on Molecular Learning Vector Quantization Neural Networks
Breast Tumor Diagnosis Based on Molecular Learning Vector Quantization Neural Networks Open
DNA nanotechnology plays a crucial role in precise cancer medicine. Currently, molecular logic circuits are applied to detect tumor‐specific biomarkers and control the release of therapeutic drugs. However, these systems lack self‐learning…
View article: Reconstruct the Pruned Model without Any Retraining
Reconstruct the Pruned Model without Any Retraining Open
Structured pruning is a promising hardware-friendly compression technique for large language models (LLMs), which is expected to be retraining-free to avoid the enormous retraining cost. This retraining-free paradigm involves (1) pruning c…
View article: MedCare: Advancing Medical LLMs through Decoupling Clinical Alignment and Knowledge Aggregation
MedCare: Advancing Medical LLMs through Decoupling Clinical Alignment and Knowledge Aggregation Open
Large language models (LLMs) have shown substantial progress in natural language understanding and generation, proving valuable especially in the medical field. Despite advancements, challenges persist due to the complexity and diversity i…
View article: Self-Localized Collaborative Perception
Self-Localized Collaborative Perception Open
Collaborative perception has garnered considerable attention due to its capacity to address several inherent challenges in single-agent perception, including occlusion and out-of-range issues. However, existing collaborative perception sys…