A5101046520- Topic Modeling
- Optical measurement and interference techniques
- Robotics and Sensor-Based Localization
- Optical Coherence Tomography Applications
- Advanced Fluorescence Microscopy Techniques
- Planetary Science and Exploration
- Astro and Planetary Science
- Structural Engineering and Vibration Analysis
- Hemophilia Treatment and Research
- Cell Image Analysis Techniques
- Space Satellite Systems and Control
- Natural Language Processing Techniques
- Risk and Safety Analysis
- Mental Health via Writing
- Explainable Artificial Intelligence (XAI)
- Text and Document Classification Technologies
- Phytoestrogen effects and research
- Robotic Path Planning Algorithms
- UAV Applications and Optimization
- Machine Learning in Healthcare
- Railway Engineering and Dynamics
- Robotic Mechanisms and Dynamics
- Image Processing Techniques and Applications
- Face recognition and analysis
- Liver Disease and Transplantation
Harbin Medical University
2024-2025
Shanghai Jiao Tong University
2022-2024
ShangHai JiAi Genetics & IVF Institute
2024
City of Hope
2024
American Thrombosis and Hemostasis Network
2024
Shanghai Artificial Intelligence Laboratory
2024
Jiangsu Institute of Metrology
2024
University of California, San Francisco
2024
Beijing General Research Institute of Mining and Metallurgy
2024
Institute of Electronics
2023
Generating realistic human motion from given action descriptions has experienced significant advancements because of the emerging requirement digital humans. While recent works have achieved impressive results in generating directly textual descriptions, they often support only a single modality control signal, which limits their application real industry. This paper presents Motion General-Purpose generaTor (MotionGPT) that can use multimodal signals, e.g., text and single-frame poses, for...
Recently, there has been growing interest in using Large Language Models (LLMs) for scientific research. Numerous benchmarks have proposed to evaluate the ability of LLMs However, current are mostly based on pre-collected objective questions. This design suffers from data leakage problem and lacks evaluation subjective Q/A ability. In this paper, we propose SciEval, a comprehensive multi-disciplinary benchmark address these issues. Based Bloom's taxonomy, SciEval covers four dimensions...
Reinforcement Learning from Human Feedback (RLHF) has become a crucial technology for aligning language models with human values and intentions, enabling to produce more helpful harmless responses. Reward are trained as proxies preferences drive reinforcement learning optimization. While reward often considered central achieving high performance, they face the following challenges in practical applications: (1) Incorrect ambiguous preference pairs dataset may hinder model accurately...
Large language models (LLMs) have formulated a blueprint for the advancement of artificial general intelligence. Its primary objective is to function as human-centric (helpful, honest, and harmless) assistant. Alignment with humans assumes paramount significance, reinforcement learning human feedback (RLHF) emerges pivotal technological paradigm underpinning this pursuit. Current technical routes usually include \textbf{reward models} measure preferences, \textbf{Proximal Policy...
Coherence scanning interferometry (CSI) is a widely used non-contact method for measuring areal surface topography. The calculation of groove depth typically employs the ‘W/3 rule’ specified in ISO 5436-1. However, batwing effect causes overshoots near edges, which can introduce noise singularities selected W/3 region calculation, thereby affecting measurement rectangular grating depth. This paper introduces definition height and width proposes simulation model that considers various...
Background Artificial intelligence (AI) and machine learning (ML) are currently used in the clinical field to improve outcome predictions on disease diagnosis prognosis. However, date, few AI/ML applications have been reported rare diseases, such as hemophilia. In this study, taking advantage of ATHNdataset, an extensive repository hemostasis thrombosis data, we aimed demonstrate application approaches build predictive models identify persons with hemophilia (PwH) who at risk poor inform...
In a depression-diagnosis-directed clinical session, doctors initiate conversation with ample emotional support that guides the patients to expose their symptoms based on diagnosis criteria. Such dialogue system is distinguished from existing single-purpose human-machine dialog systems, as it combines task-oriented and chit-chats uniqueness in topics procedures. However, due social stigma associated mental illness, data related depression consultation are rarely disclosed. Based diagnostic...
<title>Abstract</title> <bold>Background</bold> Artificial intelligence (AI) and machine learning (ML) are currently used in the clinical field to improve outcome predictions on disease diagnosis prognosis. However, date, few AI/ML applications were reported rare diseases, such as hemophilia. In this study, taking advantage of ATHNdataset, an extensive repository hemostasis thrombosis data, we aimed demonstrate application approaches build predictive models identify persons with hemophilia...
The partial label challenge in Multi-Label Class-Incremental Learning (MLCIL) arises when only the new classes are labeled during training, while past and future labels remain unavailable. This issue leads to a proliferation of false-positive errors due erroneously high confidence multi-label predictions, exacerbating catastrophic forgetting within disjoint space. In this paper, we aim refine calibration MLCIL propose Confidence Self-Calibration (CSC) approach. Firstly, for relationship...
A journal at the forefront of design and understanding solid-state crystalline materials rsc.li/crystengcommThe Royal Society Chemistry is world's leading chemistry community.Through our high impact
Background: Liver cirrhosis (LC) is one of the most significant causes morbidity and mortality in patients with chronic liver disease worldwide. Nutrition may be an important component primary prevention disease. Diet–exercise patterns frame eating behaviors exercise habits people through statistical methods related to nutritional epidemiology, which can explore relationship between living diseases among diverse populations. The purpose this study was association diet–exercise cirrhosis,...
The dynamic crack propagation trajectories play a crucial role in enhancing our understanding of spatial mechanisms involved expansion. However, visualization internal cracks under complex conditions has always been challenge. Biological networks have honed by many cycles evolutionary selection pressure and are likely to yield reasonable solutions such combinatorial optimization problems. This study applied the slime mould algorithm improve accuracy localization rocks employed Minimum...
Topographic spatial resolution is a metrological characteristic that describes the ability of surface topography measuring instrument to distinguish closely spaced features. The optical and topographic resolutions are intuitively correlative for coherence scanning interferometry (CSI), latter larger than former. This paper aims develop model study their relationships, pursue improvement through pupil modulation. First, we construct comprehensive Fourier CSI with grating taken as typical...
Topographic spatial resolution is a metrological characteristic that describes the ability of surface topography measuring instrument to distinguish closely spaced features. The optical and topographic resolutions are intuitively correlative for coherence scanning interferometry (CSI), latter larger than former. This paper aims develop model study their relationships, pursue improvement through pupil modulation. First, we construct comprehensive Fourier CSI with grating taken as typical...
Coherence scanning interferometry (CSI) is a non-destructive method for measuring the microstructure surface topography, but it fails to retrieve bottom topography because detection light blocked by sidewalls of high aspect ratio (HAR) samples. Our team has proposed CSI technology with transparent sample measure thus ensuring numerical aperture throughput. However, dedicated optical path monitor aberrations caused modulation from necessary and complex added aberration correction, which...
This study aims to establish an artificial intelligence model, ThyroidNet, diagnose thyroid nodules using deep learning techniques accurately. A novel method, is introduced and evaluated based on for the localization classification of nodules. First, we propose multitask TransUnet, which combines TransUnet encoder decoder with learning. Second, DualLoss function, tailored nodule tasks. It balances tasks help improve model's generalization ability. Third, introduce strategies augmenting data....