A5013071579- Refrigeration and Air Conditioning Technologies
- Infrastructure Maintenance and Monitoring
- Asphalt Pavement Performance Evaluation
- Advanced Sensor Technologies Research
- Privacy-Preserving Technologies in Data
- Advanced Thermodynamic Systems and Engines
- Freezing and Crystallization Processes
- Hydraulic and Pneumatic Systems
- Magnetic and Electromagnetic Effects
- Misinformation and Its Impacts
- Ethics and Social Impacts of AI
- Fault Detection and Control Systems
- Flame retardant materials and properties
- Transport Systems and Technology
- Sirtuins and Resveratrol in Medicine
- Proteoglycans and glycosaminoglycans research
- Concrete Corrosion and Durability
- Adversarial Robustness in Machine Learning
- Digital Media Forensic Detection
- Domain Adaptation and Few-Shot Learning
- Dielectric materials and actuators
- Image and Video Stabilization
- Synthesis and properties of polymers
- Cancer-related molecular mechanisms research
- Advanced ceramic materials synthesis
University of Technology Sydney
2023-2024
Chongqing Jiaotong University
2012-2022
Zhejiang University
2019
Xi'an Jiaotong University
2015-2017
Second Affiliated Hospital of Zhejiang University
2014-2015
Blood Center of Zhejiang Province
2015
China Institute of Atomic Energy
2009
Abstract Hypoxia preconditioning enhances the therapeutic effect of mesenchymal stem cells (MSCs). However, mechanism underlying hypoxia-induced augmentation protective MSCs on myocardial infarction (MI) is poorly understood. We show that hypoxia-enhanced survival, mobility, and protection cocultured cardiomyocytes were paralleled by increased expression leptin cell surface receptor CXCR4. The enhanced activities abolished either knockdown with a selective shRNA or genetic deficiency its in...
Federated Learning (FL) has gained significant attention as it facilitates collaborative machine learning among multiple clients without centralizing their data on a server. FL ensures the privacy of participating by locally storing data, which creates new challenges in fairness. Traditional debiasing methods assume centralized access to sensitive information, rendering them impractical for setting. Additionally, is more susceptible fairness issues than due diverse client sources that may be...
Machine unlearning aims to enable models forget specific data instances when receiving deletion requests. Current research centers on efficient erase the influence of from model and neglects subsequent impacts remaining data. Consequently, existing algorithms degrade model's performance after unlearning, known as over-unlearning. This paper addresses this critical yet under-explored issue by introducing machine Unlearning via Null Space Calibration (UNSC), which can accurately unlearn target...
Supersonic ejectors involve very complex phenomena such as interaction between supersonic and subsonic flows, shock trains, instabilities, which strongly influences the performance of ejector. In this study, static pressure distribution along ejector wall Mach number axis are used to investigate internal flow field Results indicate that when back is much less than critical pressure, there two series change will not affect before effective area section, so entrainment ratio would remain...
The crystalline blockage of tunnel drainage pipes in a karst area seriously affects the normal operation system and buries hidden dangers for tunnel. In order to obtain influencing factors laws pipe crystallization area, based on field investigation plugging, effects groundwater velocity, diameter, material, structure law are studied by indoor model test. results show that: (1) With increase diameter (20–32 mm), crystallinity first increases then decreases. (2) water velocity (22.0–63.5...
Faulting prediction is the core of concrete pavement maintenance and design. Highway agencies are always faced with problem lower accuracy for which causes costly maintenance. Although many researchers have developed some performance models, has remained a challenge. This paper reviews models JPCP faulting that been used in past research. Then three including multivariate nonlinear regression (MNLR) model, artificial neural network (ANN) Markov Chain (MC) model tested compared using set...
Abstract A new 1D model using the real gas property is proposed to predict ejector performance at critical and sub-critical operational modes, while most previous models usually used ideal only predicted mode operation. Constant pressure mixing assumed occur inside constant area section of operation, effectiveness verified against experimental data. The results show that accurately predicts over all ranges a useful tool for predicting within larger refrigeration cycle models.
In order to explore the aging process and mechanism of new old asphalt in plant hot-mix recycled mixture during mixing use process, liquid viscosity test low-temperature bending creep are carried out dynamic viscosity, stiffness modulus, rate, flexibility matrix asphalt, after rolling thin film oven (RTFOT) pressurized vessel (PAV) aging. The macroscopic performance attenuation law thermal regeneration is compared analyzed. After that, explored by infrared spectroscopy differential...
Machine unlearning aims to enable models forget specific data instances when receiving deletion requests. Current research centres on efficient erase the influence of from model and neglects subsequent impacts remaining data. Consequently, existing algorithms degrade model's performance after unlearning, known as \textit{over-unlearning}. This paper addresses this critical yet under-explored issue by introducing machine \underline{U}nlearning via \underline{N}ull \underline{S}pace...
Isolated Sign Language Recognition (ISLR) focuses on identifying individual sign language glosses. Considering the diversity of languages across geographical regions, developing region-specific ISLR datasets is crucial for supporting communication and research. Auslan, as a specific to Australia, still lacks dedicated large-scale word-level dataset task. To fill this gap, we curate \underline{\textbf{the first}} Multi-view Multi-modal Word-Level Australian recognition dataset, dubbed...
Transfer learning is an important approach that produces pre-trained teacher models which can be used to quickly build specialized student models. However, recent research on transfer has found it vulnerable various attacks, e.g., misclassification and backdoor attacks. still not clear whether model inversion Launching a attack against scheme challenging. Not only does the hide its structural parameters, but also inaccessible adversary. Hence, when targeting model, both white-box black-box...
The second and third types of boundary conditions were mixed to simulate the change rule for subgrade temperature field under actual climate condition in a seasonal frozen region. difference between sunny shady slopes was also analyzed. freezing thawing coefficients introduced compute deformation caused by differences considering location shape fringe. Then, mechanism discussed. Results show following conditions. (1) When groundwater exists, an uneven horizontal is strongly affected slopes....