A5101123756- Natural Language Processing Techniques
- Topic Modeling
- Trigeminal Neuralgia and Treatments
- Advanced Neuroimaging Techniques and Applications
- MRI in cancer diagnosis
- Functional Brain Connectivity Studies
- Advanced MRI Techniques and Applications
- Text Readability and Simplification
- Lanthanide and Transition Metal Complexes
- Migraine and Headache Studies
- Machine Learning and Data Classification
- Radiomics and Machine Learning in Medical Imaging
- Speech Recognition and Synthesis
- Sleep and Wakefulness Research
- Microstructure and Mechanical Properties of Steels
- Cerebrovascular and Carotid Artery Diseases
- Hydrogen embrittlement and corrosion behaviors in metals
- Multimodal Machine Learning Applications
- EEG and Brain-Computer Interfaces
- Software Engineering Research
- Welding Techniques and Residual Stresses
- Pain Mechanisms and Treatments
- Medical Imaging Techniques and Applications
- Speech and dialogue systems
- Vascular Malformations Diagnosis and Treatment
Jinan University
2024
Southern Medical University
2010-2024
Guangdong Provincial People's Hospital
2015-2024
University of Science and Technology Beijing
2012-2024
Tencent (China)
2022-2023
Westlake University
2023
Zhejiang University
2023
Zhujiang Hospital
2010
Purpose Insomnia is the second most common mental disorder. However, topologic alterations in structural brain connectome patients with primary insomnia (PI) remain largely unknown. Methods A total of 44 PI and 46 age-, gender- education level matched healthy control (HC) participants were recruited this study. Diffusion tensor imaging (DTI) resting state MRI used to construct for each participant, network parameters employed by nonparametric permutations evaluate significant differences...
Neuroimaging studies have documented brain structural alterations induced by chronic pain, particularly in gray matter volume. However, the effects of trigeminal neuralgia (TN), a severe paroxysmal pain disorder, on cortical morphology are not yet known. In this study, we recruited 30 TN patients and age-, gender-matched healthy controls (HCs). Using Computational Anatomy Toolbox (CAT12), calculated compared group differences thickness, gyrification, sulcal depth with two-sample t tests (p <...
Accumulating evidence from neuroimaging studies has supported that chronic pain could induce changes in brain function. However, few have focused on the dynamic regional homogeneity (dReHo) of trigeminal neuralgia (TN). In this study, twenty-eight TN patients and 28 healthy controls (HC) were included. Based resting-state fMRI (rsfMRI), we detected abnormalities dReHo patients. Patients with had decreased left middle temporal gyrus, superior parietal lobule, precentral increased thalamus....
Abstract Background Neurovascular contact (NVC) is the main cause of primary trigeminal neuralgia (PTN); however, cases PTN without NVC are still observed. In this study, Meckel cave (MC) morphology in were analyzed by radiomics and compared to healthy controls (HCs) explore PTN. Methods We studied 3.0T MRI data 115 patients with 46 HCs. Bilateral MC was modeled using 3D Slicer software, morphological characteristics method. Results The right side incidence rate group higher than left...
While large-scale neural language models, such as GPT2 and BART, have achieved impressive results on various text generation tasks, they tend to get stuck in undesirable sentence-level loops with maximization-based decoding algorithms (\textit{e.g.}, greedy search). This phenomenon is counter-intuitive since there are few consecutive repetitions human corpora (e.g., 0.02\% Wikitext-103). To investigate the underlying reasons for generating repetitions, we study relationship between...
Early neurologic deterioration occurs in up to one-third of patients with acute ischemic stroke (IS), often leading poor functional outcomes. At present, few studies have applied amide proton transfer (APT) imaging the evaluation early neurological (END). This study analyzed value computed tomography perfusion (CTP) combined multimodal magnetic resonance (MRI) IS END.
The impressive performance of Large Language Models (LLMs) has consistently surpassed numerous human-designed benchmarks, presenting new challenges in assessing the shortcomings LLMs. Designing tasks and finding LLMs' limitations are becoming increasingly important. In this paper, we investigate question whether an LLM can discover its own from errors it makes. To end, propose a Self-Challenge evaluation framework with human-in-the-loop. Starting seed instances that GPT-4 fails to answer,...
Background: End-stage renal disease (ESRD) is a serious public health problem, which can often lead to multiorgan dysfunction, such as cerebrovascular and cognitive damage. It essential understand impairment in patients with ESRD develop better treatment prevent further impairment. Cognitive believed be related structural abnormalities the brain. Purpose: To investigate white matter microstructural using TBSS analysis of DTI explore possible mechanisms underlying impaired function. Materials...
Background: Pre-operative non-invasive differentiation of benign and malignant thyroid nodules is difficult for doctors. This study aims to determine whether amide proton transfer (APT) imaging zonally oblique multi-slice (ZOOM) diffusion-weighted (DWI) can provide increased accuracy in differentiating nodules. Methods: retrospective was approved by the institutional review board included 60 50 patients. All were classified as (n = 21) or 39) based on pathology. It meaningful analyze APT...
This paper introduces WeChat AI's participation in WMT 2021 shared news translation task on English->Chinese, English->Japanese, Japanese->English and English->German. Our systems are based the Transformer (Vaswani et al., 2017) with several novel effective variants. In our experiments, we employ data filtering, large-scale synthetic generation (i.e., back-translation, knowledge distillation, forward-translation, iterative in-domain transfer), advanced finetuning approaches,...
This study comprehensively evaluates the translation quality of Large Language Models (LLMs), specifically GPT-4, against human translators varying expertise levels across multiple language pairs and domains. Through carefully designed annotation rounds, we find that GPT-4 performs comparably to junior in terms total errors made but lags behind medium senior translators. We also observe imbalanced performance different languages domains, with GPT-4's capability gradually weakening from...
Yulong Chen, Huajian Zhang, Yijie Zhou, Xuefeng Bai, Yueguan Wang, Ming Zhong, Jianhao Yan, Yafu Li, Judy Xianchao Zhu, Yue Zhang. Proceedings of the 61st Annual Meeting Association for Computational Linguistics (Volume 1: Long Papers). 2023.
Most existing text generation models follow the sequence-to-sequence paradigm. Generative Grammar suggests that humans generate natural language texts by learning grammar. We propose a syntax-guided schema, which generates sequence guided constituency parse tree in top-down direction. The decoding process can be decomposed into two parts: (1) predicting infilling for each constituent lexicalized syntax context given source sentence; (2) mapping and expanding to construct next-level context....
Accumulating evidence suggests that trigeminal neuralgia (TN) causes structural and functional alterations in the brain. However, only a few studies have focused on cerebral blood flow (CBF) changes patients with TN. This study aimed to explore whether altered perfusion patterns exist TN investigate relationship between abnormal regional CBF (rCBF) clinical characteristics of TN.This included 28 30 age- sex-matched healthy controls (HCs) who underwent MRI (fMRI) brain using pseudo-continuous...
Large language models (LLMs) trained on vast corpora suffer from inevitable stereotype biases. Mitigating these biases with fine-tuning could be both costly and data-hungry. Model editing methods, which focus modifying LLMs in a post-hoc manner, are of great potential to address debiasing. However, it lacks comprehensive study that facilitates internal external model supports various bias types, as well understands the pros cons applying methods stereotypical To mitigate this gap, we...
The application scope of large language models (LLMs) is increasingly expanding. In practical use, users might provide feedback based on the model's output, hoping for a responsive model that can complete responses according to their feedback. Whether appropriately respond users' refuting and consistently follow through with execution has not been thoroughly analyzed. light this, this paper proposes comprehensive benchmark, RefuteBench, covering tasks such as question answering, machine...