- Toxic Organic Pollutants Impact
- Privacy-Preserving Technologies in Data
- Stochastic Gradient Optimization Techniques
- Microplastics and Plastic Pollution
- Advanced Graph Neural Networks
- Advanced Vision and Imaging
- Advanced Image Processing Techniques
- Pesticide Exposure and Toxicity
- Parasitic infections in humans and animals
- Effects and risks of endocrine disrupting chemicals
- Pesticide and Herbicide Environmental Studies
- Image Enhancement Techniques
- Topic Modeling
- Data Visualization and Analytics
- Advanced Neural Network Applications
- Domain Adaptation and Few-Shot Learning
- Big Data Technologies and Applications
- Acupuncture Treatment Research Studies
- Time Series Analysis and Forecasting
- Environmental Toxicology and Ecotoxicology
- Natural Language Processing Techniques
- Recommender Systems and Techniques
- Cryptography and Data Security
- Artificial Intelligence in Healthcare
- Machine Learning in Healthcare
Tianjin University of Technology
2024-2025
Guangdong University of Petrochemical Technology
2022-2024
Adobe Systems (United States)
2024
University of Science and Technology of China
2023-2024
Shenzhen University Health Science Center
2022
University of British Columbia
2022
Kunming Municipal Hospital of Traditional Chinese Medicine
2021
Universiti Sains Malaysia
2020
With the development of artificial intelligence technology, more and products are being applied to education industry. Many countries in world have also formulated relevant policies promote application technology education. This paper briefly discusses history its field including teaching learning innovations, effective approaches smart campus life styles. research analyzes changes brought by from different aspects. It is suggested that, order better education, there three important aspects...
Vertical Federated Learning (VFL) enables multiple data owners, each holding a different subset of features about largely overlapping sets sample(s), to jointly train useful global model. Feature selection (FS) is important VFL. It still an open research problem as existing FS works designed for VFL either assumes prior knowledge on the number noisy or post-training threshold be selected, making them unsuitable practical applications. To bridge this gap, we propose Stochastic Dual-Gate based...
To reveal the pollution status of agricultural soils along with rapid urbanization and economic growth, a large regional survey organochlorine pesticides (OCPs) in was conducted Pearl River Delta (PRD) China. The results showed that total residues 23 OCPs were range ND-946 ng/g dry weight. OCP distinct spatial distribution characteristics within PRD. mainly found areas high production industrial activities. Higher concentrations observed top layer soil, while concentration decreases to...
Phthalate esters (PAEs) are widely used as plasticizers in industrial and commercial products, classified endocrine-disrupting compounds. In this study, we investigated the contamination characteristics health risks of PAEs soil-plant system coastal areas South China. were detected soil plant samples at all 37 sampling sites. The total concentration 15 ranged from 0.445 to 4.437 mg/kg, mean was 1.582 ± 0.937 mg/kg. 2.176 30.276 8.712 5.840 Di(2-Ethylhexyl) phthalate (DEHP) di-n-butyl (DnBP)...
Organophosphate esters (OPEs) are frequently used as flame retardants and plasticizers in various commercial products. While initially considered substitutes for brominated retardants, they have faced restrictions some countries due to their toxic effects on organisms. We collected 37 soil crop samples 20 cities along the coast of South China, OPEs were detected all them. Meanwhile, we studied contamination potential human health risks OPEs. In samples, combined concentrations eight varied...
In order to investigate the pollution status of polycyclic aromatic hydrocarbons (PAHs) in agricultural soil, 240 soil topsoil samples were collected from nine Pearl River Delta cities June September 2019. addition, 72 for vertical profiles, which profiles excavated a depth 80 cm. After sample preparation, GC-MS was used separation compounds on HP-5MS quartz capillary column. ArcGIS software map spatial distribution. Health risk assessment conducted using USEPA standard. The results showed...
Vertical Federated Learning (VFL) enables multiple data owners, each holding a different subset of features about largely overlapping set samples, to collaboratively train global model. The quality owners' local affects the performance VFL model, which makes feature selection vitally important. However, existing methods for either assume availability prior knowledge on number noisy or post-training threshold useful be selected, making them unsuitable practical applications. To bridge this...
Conversion rate (CVR) prediction is essential in recommender systems, facilitating precise matching between recommended items and users' preferences. However, the sample selection bias (SSB) data sparsity (DS) issues pose challenges to accurate prediction. Existing works have proposed click-through conversion (CTCVR) task which models samples from exposure ``click conversion" entire space incorporates multi-task learning. This approach has shown efficacy mitigating these challenges....
Vertical Federated Learning (VFL) enables multiple data owners, each holding a different subset of features about largely overlapping sets sample(s), to jointly train useful global model. Feature selection (FS) is important VFL. It still an open research problem as existing FS works designed for VFL either assumes prior knowledge on the number noisy or post-training threshold be selected, making them unsuitable practical applications. To bridge this gap, we propose Stochastic Dual-Gate based...
In this paper, we introduce an approach for future frames prediction based on a single input image. Our method is able to generate entire video sequence the information contained in frame. We adopt autoregressive our generation process, i.e., output from each time step fed as next step. Unlike other methods that use "one shot" generation, preserve much more details image, while also capturing critical pixel-level changes between frames. overcome problem of quality degradation by introducing...
In this work, we revisit linguistic acceptability in the context of large language models. We introduce CoLAC - Corpus Linguistic Acceptability Chinese, first large-scale dataset for a non-Indo-European language. It is verified by native speakers and that comes with two sets labels: linguist label crowd label. Our experiments show even largest InstructGPT model performs only at chance level on CoLAC, while ChatGPT's performance (48.30 MCC) also much below supervised models (59.03 human...
In this paper, we tackle an open research question in transfer learning, which is selecting a model initialization to achieve high performance on new task, given several pre-trained models. We propose highly efficient and accurate approach based duality diagram similarity (DDS) between deep neural networks (DNNs). DDS generic framework represent compare data of different feature dimensions. validate our the Taskonomy dataset by measuring correspondence actual learning rankings 17 taskonomy...
Despite the abundance of subphenotype clustering studies on sepsis and acute kidney injury (AKI), few models consider real-time information clinical features. The lack supervision may lead to patient subgroups being derived as clusters without stratification patients based outcome interests. sensitivity dimension in methods is generally ignored, so robustness. In this study, we propose an ensembled outcome-driven bidirectional long short-term memory autoencoder (BiLSTM-AE) architecture with...