A5102734597- Advanced Graph Neural Networks
- Topic Modeling
- Recommender Systems and Techniques
- Complex Network Analysis Techniques
- Sentiment Analysis and Opinion Mining
- IoT and Edge/Fog Computing
- Digital Imaging for Blood Diseases
- Spam and Phishing Detection
- Domain Adaptation and Few-Shot Learning
- Machine Learning and Algorithms
- Energy, Environment, and Transportation Policies
- Privacy-Preserving Technologies in Data
- Diabetic Foot Ulcer Assessment and Management
- Natural Language Processing Techniques
- Interstitial Lung Diseases and Idiopathic Pulmonary Fibrosis
- Global Energy Security and Policy
- Biomedical Text Mining and Ontologies
- Blockchain Technology Applications and Security
- COVID-19 Clinical Research Studies
- Energy, Environment, Economic Growth
- Video Surveillance and Tracking Methods
- Caching and Content Delivery
- Human Mobility and Location-Based Analysis
- COVID-19 and healthcare impacts
- Machine Learning in Healthcare
Inner Mongolia University
2025
Aerospace Center Hospital
2025
University of Illinois Chicago
2020-2024
Nanjing Medical University
2020
Detecting hot social events (e.g., political scandal, momentous meetings, natural hazards, etc.) from messages is crucial as it highlights significant happenings to help people understand the real world. On account of streaming nature messages, incremental event detection models in acquiring, preserving, and updating over time have attracted great attention. However, challenge that existing methods towards are generally confronted with ambiguous features, dispersive text contents, multiple...
GNNs have been widely used in deep learning on graphs. They learn effective node representations. However, most methods ignore the heterogeneity. Methods designed for heterogeneous graphs, other hand, fail to complex semantic representations because they only use meta-paths instead of meta-graphs. Furthermore, cannot fully capture content-based correlations, as either do not self-attention mechanism or it consider immediate neighbors each node, ignoring higher-order neighbors. We propose a...
Recommendation systems suffer in the strict cold-start (SCS) scenario, where user-item interactions are entirely unavailable. The well-established, dominating identity (ID)-based approaches completely fail to work. Cold-start recommenders, on other hand, leverage item contents (brand, title, descriptions, etc.) map new items existing ones. However, SCS recommenders explore coarse-grained manners that introduce noise or information loss. Moreover, informative data sources than contents, such...
Social events provide valuable insights into group social behaviors and public concerns therefore have many applications in fields such as product recommendation crisis management. The complexity streaming nature of messages make it appealing to address event detection an incremental learning setting, where acquiring, preserving, extending knowledge are major concerns. Most existing methods, including those based on clustering community detection, learn limited amounts they ignore the rich...
With the development of Vehicular Edge Computing (VEC) computing architectures, in study task offloading problem, based on differences delay sensitivity and dynamic characteristics environmental information, this paper designs a distributed DRL framework non-cooperative game, introduces memory mechanism(RNN) shared experience mechanism (Shared Experience Actor-Critic) MADDPG algorithm, which improves learning efficiency algorithm as well convergence speed by capturing sharing timeseries data...
Patients with non-Hodgkin lymphoma (NHL) face heightened mortality and accelerated disease progression when persistently infected severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). This critical situation underscores the urgent need to identify risk factors establish early intervention strategies tailored this vulnerable population. The primary aim of study was investigate associated persistent SARS-CoV-2 infection in NHL patients during COVID-19 pandemic. A retrospective cohort...
Temporal Expression Extraction (TEE) is essential for understanding time in natural language. It has applications Natural Language Processing (NLP) tasks such as question answering, information retrieval, and causal inference. To date, work this area mostly focused on English there a scarcity of labeled data other languages. We propose XLTime, novel framework multilingual TEE. XLTime works top pre-trained language models leverages multi-task learning to prompt cross-language knowledge...
Mobile object tracking, which has broad applications, utilizes a large number of Internet Things (IoT) devices to identify, record, and share the trajectory information physical objects. Nonetheless, IoT are energy constrained not feasible for deploying advanced tracking techniques due significant computing requirements. To address these issues, in this paper, we develop an edge computing-based multivariate time series (EC-MTS) framework accurately track mobile objects exploit offload its...
State-of-the-art Graph Neural Networks (GNNs) have achieved tremendous success in social event detection tasks when restricted to a closed set of events. However, considering the large amount data needed for training and limited ability neural network handling previously unknown data, it is hard existing GNN-based methods operate an open setting. To address this problem, we design Quality-aware Self-improving Network (QSGNN) which extends knowledge from known by leveraging best samples...
Objective To discuss the application of artificial intelligence automatic diatom identification system in practical cases, to provide reference for quantitative analysis using and validate deep learning model incorporated into system. Methods Organs from 10 corpses water were collected digested with nitric acid; then smears digitally scanned a digital slide scanner diatoms tested qualitatively quantitatively by Results The area under curve （AUC） receiver operator characteristic （ROC） system,...
Graph neural networks (GNNs) have been widely used in deep learning on graphs. They can learn effective node representations that achieve superior performances graph analysis tasks such as classification and clustering. However, most methods ignore the heterogeneity real-world Methods designed for heterogeneous graphs, other hand, fail to complex semantic because they only use meta-paths instead of meta-graphs. Furthermore, cannot fully capture content-based correlations between nodes,...
Recommender systems (RecSys) aim to predict users' preferences based on historical interactions and content profiles, they are vital components of many online services. However, the strict cold-start (SCS) issue, i.e., users/items have no prior interactions, poses significant challenges for RecSys. The existing methods seek transfer knowledge, collaborative filtering (CF) or combine two from warm-start scenario towards (strict) scenarios. these approaches either ignore available information...