A5101617928- Domain Adaptation and Few-Shot Learning
- Electrical Contact Performance and Analysis
- Advanced Neural Network Applications
- Aerospace Engineering and Energy Systems
- Topic Modeling
- Viral gastroenteritis research and epidemiology
- Multimodal Machine Learning Applications
- Advanced Image and Video Retrieval Techniques
- Animal Virus Infections Studies
- Blind Source Separation Techniques
- Machine Fault Diagnosis Techniques
- Extracellular vesicles in disease
- Advanced materials and composites
- Natural Language Processing Techniques
- HIV Research and Treatment
- Mechanical stress and fatigue analysis
- Simulation and Modeling Applications
- Intermetallics and Advanced Alloy Properties
- Mental Health via Writing
- Cavitation Phenomena in Pumps
- Aesthetic Perception and Analysis
- Advanced Graph Neural Networks
- EEG and Brain-Computer Interfaces
- Visual Attention and Saliency Detection
- Acute Myeloid Leukemia Research
Soochow University
2024-2025
Second Affiliated Hospital of Soochow University
2024-2025
Chinese University of Hong Kong
2023-2024
Guilin University of Electronic Technology
2024
University of Electronic Science and Technology of China
2024
Bioscience (China)
2024
Southwest University
2024
Second Affiliated Hospital of Zhejiang University
2024
Beijing Sport University
2024
Xihua University
2024
Diffuse large B-cell lymphoma (DLBCL) is characterized by its aggressive nature and resistance to standard chemotherapy, necessitating the development of new therapeutic approaches. The emergence natural products their derivatives has notably influenced cancer treatment, making morusinol, a medicine-derived monomer, promising candidate. Here, we showed that morusinol exerted antitumor effects on DLBCL in vitro inducing apoptosis cell cycle arrest. Impressively, treatment exhibited potent...
High-dose methotrexate (HD-MTX) serves as the cornerstone of central nervous system lymphoma (CNSL) treatment, but its efficacy is limited due to low blood-brain barrier (BBB) penetration and adverse effects. This study focused on an exosome-based drug delivery approach aimed at enhancing BBB permeability, thereby reducing required dosage (MTX) while ensuring specific targeting CNSL. Human adipose-derived mesenchymal stem cells (hAMSCs) were modified with a lentiviral vector encoding...
Exosomes have emerged as promising vehicles for delivering therapeutic cargoes to specific cells or tissues, owing their superior biocompatibility, reduced immunogenicity, and enhanced targeting capabilities compared with conventional drug delivery systems. In this study, we developed a platform using exosomes derived from monocytes, specifically designed targeted of bortezomib (Btz) multiple myeloma (MM) cells. Our approach involved the genetic modification monocytes express antibodies...
Nontypeable Haemophilus influenzae (NTHi) is a major cause of community acquired pneumonia and exacerbation chronic obstructive pulmonary disease. A current effort in NTHi vaccine development has focused on generating humoral responses been greatly impeded by antigenic variation among the numerous circulating strains. In this study, we showed that immunization mice with killed generated broad protection against lung infection different While passive transfer immune antibodies protected only...
The progress of deep learning technology has made image classification an important application field. Image style is a complex task involving the recognition whole picture, including salient features and detailed features. This study based on ResNet algorithm improved its Resnet 50 version with excellent performance. In model architecture, we introduce blur pool operation replace traditional Relu function Celu activation function. addition, triplet attention mechanism was integrated to...
Fretting, which induced by high frequency vibration, remains a major cause of connector failure and can impair reliability in aerospace electrical electronic systems. The test system was developed for fretting conditions (Maximum vibration 2000Hz) using electro dynamic shaker this paper. behavior gold plated copper contacts studied controlling the amplitude. Contact resistance, friction force, displacement were recorded explicitly. Surface characteristics fretted area are also analyzed...
The object inference for augmented reality (AR) requires a precise localization within user's physical environment and the adaptability to dynamic communication conditions. Deep learning (DL) is advantageous in capturing highly-nonlinear features of diverse data sources drawn from complex objects. However, existing DL techniques may have disfluency or instability issues when deployed on resource-constrained devices with poor conditions, resulting bad user experiences. This article addresses...
During the coronavirus disease 2019 (COVID-19) pandemic, early detection and isolation of individuals infected with severe acute respiratory syndrome 2 (SARS-CoV-2) through mass testing can effectively prevent transmission. SARS-CoV-2 nucleic acid rapid based on loop-mediated isothermal amplification (LAMP) may be appropriate to include in procedures. We used 860 nasopharyngeal specimens from healthcare workers Huashan Hospital COVID-19 patients collected April 7th 21st, 2022, assess...
Knowledge distillation aims at transferring knowledge from the teacher model to student one by aligning their distributions. Feature-level often uses L2 distance or its variants as loss function, based on assumption that outputs follow normal This poses a significant challenge when distribution gaps are substantial since this function ignores variance term. To address problem, we propose decompose transfer objective into small parts and optimize it progressively. process is inspired...
Abstract Aimed at the problem that quality of formation aero-rotor blades with complex shaped structures is difficult to predict due coupling each physical field in process electrochemical machining, a mathematical model machining characterized by conductivity multi-physics established firstly studying action mechanism flow and temperature on electric process. Then, utilizing Runge–Kutta method, relationship between machine path analysed; Finally, based this relationship, erosion compared...
Many work and living activities in modern society are increasingly dependent on web app software. However, much existing software lacks consideration of many diverse end-user characteristics, such as age, mental physical challenges, language proficiency, culture, socio-economic status, educational attainment so on. developers do not have lived experience most these fail to empathise with those who them, lack knowledge how address them their design evaluation activities. To this issue, we...
It is an open problem to compile DNN models on GPU and improve the performance. A novel framework, GTuner, proposed jointly learn from structures of computational graphs statistical features codes find optimal code implementations. Graph ATtention network (GAT) designed as performance estimator in GTuner. In GAT, graph neural layers are used propagate information a multi-head self-attention module complicated relationships between features. Under guidance generated through auto-tuning....
Space-filling curves (SFCs) act as a linearization approach to map data in higher dimensional space lower space, which is used comprehensively computer vision, such image/point cloud compression, hashing and etc. Currently, researchers formulate the problem of searching for an optimal SFC finding single Hamiltonian circuit on image grid graph. Existing methods adopt graph neural networks (GNN) search. By modeling pixel graph, they first GNN predict edge weights then generate minimum spanning...
The advancement of mobile edge computing (MEC) is driving the utilization devices (MDs) for real-time video detection. Nonetheless, during emergency scenarios, detection tasks encounter two primary challenges: 1) time-varying channels over time between MDs and unmanned aerial vehicle (UAV); 2) need precision while managing energy consumption on MDs. In this paper, we collaboratively address optimization computation offloading, data compression, resource allocation challenges in context...
As one of the important development direction future aircraft, stratospheric airship would meet demand diurnal cycle ultra-long duration flight. So energy storage system must be managed reliably. Factors will researched and analyzed which influence battery management (BMS) airship. The main functions software for are included as follows: estimation state charge, data acquisition, thermal management, safety communication. This research could provide effective reference specific application...
Objective Detailed studies of correlation between HIV-M.tb co-infection and hierarchy declines CD8+/CD4+ T-cell counts IFN-γ responses have not been done. We conducted case-control to address this issue. Methods 164 HIV-1-infected individuals comprised HIV-1+ATB, HIV-1+LTB HIV-1+TB- groups were evaluated. Immune phenotyping complete blood count (CBC) employed measure CD4+ CD8+ counts; T.SPOT.TB intracellular cytokine staining (ICS) utilized detect ESAT6, CFP10 or PPD-specific responses....
Deep neural networks (DNNs) have been widely used recently while their hardware deployment optimizations are very time-consuming and the historical knowledge is not utilized efficiently. In this paper, to accelerate optimization process find better configurations, we propose a novel transfer learning method based on deep Gaussian processes (DGPs). Firstly, (DGP) model built data learn empirical knowledge. Secondly, new task, tuning set sampled for task under guidance of DGP model. Then tuned...