A5017536603- Advanced Graph Neural Networks
- Complex Network Analysis Techniques
- Anomaly Detection Techniques and Applications
- Functional Brain Connectivity Studies
- Network Security and Intrusion Detection
- Brain Tumor Detection and Classification
- EEG and Brain-Computer Interfaces
- Vehicle emissions and performance
- Advanced Computing and Algorithms
- Intelligent Tutoring Systems and Adaptive Learning
- Text and Document Classification Technologies
- Air Quality and Health Impacts
- Computational Drug Discovery Methods
- Air Quality Monitoring and Forecasting
- Neural Networks and Applications
- Domain Adaptation and Few-Shot Learning
- Educational Technology and Assessment
- Data-Driven Disease Surveillance
- Graph Theory and Algorithms
- Adversarial Robustness in Machine Learning
- Online Learning and Analytics
- Dementia and Cognitive Impairment Research
- Machine Learning in Healthcare
- Advanced Neural Network Applications
- Student Assessment and Feedback
Macquarie University
2020-2024
Tianjin University of Science and Technology
2020-2021
Graph anomaly detection, here, aims to find rare patterns that are significantly different from other nodes. Attributed graphs containing complex structure and attribute information ubiquitous in our life scenarios such as bank account transaction graph paper citation graph. Anomalous nodes on attributed show great difference others the perspectives of attributes, give rise various types anomalies. In this paper, we investigate three anomalies: local, global, And, neural networks (GNNs)...
Abstract Graph level anomaly detection (GLAD) aims to spot anomalous graphs that structure pattern and feature information are different from most normal in a graph set, which is rarely studied by other researchers but has significant application value. For instance, GLAD can be used distinguish some characteristic molecules drug discovery chemical analysis. However, mainly faces the following three challenges: (1) learning more comprehensive representations differ abnormal graphs, (2)...
The human brain is a highly complex neurological system that has been the subject of continuous exploration by scientists. With help modern neuroimaging techniques, there significant progress made in disorder analysis. There an increasing interest about utilizing artificial intelligence techniques to improve efficiency diagnosis recent years. However, these methods rely only on data for and do not explore pathogenic mechanism behind or provide interpretable result toward decision....
In this paper, we propose an interpretable brain graph contrastive learning framework, which aims to learn representations by a unsupervised way for disorder prediction and pathogenic analysis. Our framework consists of two key designs: We first utilize the controllable data augmentation strategy perturb unimportant structures attribute features generation graphs. Then, considering that difference healthy patient graphs is small, introduce hard negative sample evaluation weight samples loss,...
Exploring the complex structure of human brain is crucial for understanding its functionality and diagnosing disorders. Thanks to advancements in neuroimaging technology, a novel approach has emerged that involves modeling as graph-structured pattern, with different regions represented nodes functional relationships among these edges. Moreover, graph neural networks (GNNs) have demonstrated significant advantage mining data. Developing GNNs learn representations disorder analysis recently...
Network representation aims to learn low-dimensional vector representations of network nodes while preserving the inherent properties network. For all its popularity, majority existing methods focus on exploitation diverse information, including topology and semantic information network, ignore their implicit semantics. example, we know saying that birds a feather flock together. More concretely, one node can be influenced by neighbors' information. Furthermore, even two are not directly...
Graph structure patterns are widely used to model different area data recently. How detect anomalous graph information on these has become a popular research problem. The objective of this is centered the particular issue that how abnormal graphs within set. previous works have observed mainly show node-level and graph-level anomalies, but methods equally treat two anomaly forms above in evaluation graphs, which contrary fact types degrees terms anomalies. Furthermore, subtle differences...
Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI
Exploring the complex structure of human brain is crucial for understanding its functionality and diagnosing disorders. Thanks to advancements in neuroimaging technology, a novel approach has emerged that involves modeling as graph-structured pattern, with different regions represented nodes functional relationships among these edges. Moreover, graph neural networks (GNNs) have demonstrated significant advantage mining data. Developing GNNs learn representations disorder analysis recently...
Graph-level anomaly detection (GLAD) has already gained significant importance and become a popular field of study, attracting considerable attention across numerous downstream works. The core focus this domain is to capture highlight the anomalous information within given graph datasets. In most existing studies, anomalies are often instances few. stark imbalance misleads current GLAD methods on learning patterns normal graphs more, further impacting performance. Moreover, predominantly...
In order to analyse the real-world information network more effectively, we propose a hierarchical embedding method based on partitioning, NPHNE. NPHNE is nested algorithm that can be combined with existing baseline algorithms enhance their representation. consists of two parts: graph abstracting and propagation. The process as follows: Firstly, modularity used pre-determine partition, purpose constrain maximum number levels. Then, hybrid collapsing method, series abstract graphs...
To reduce the repetitive and complex work of instructors, exam paper generation (EPG) technique has become a salient topic in intelligent education field, which targets at generating high-quality automatically according to instructor-specified assessment criteria. The current advances utilize ability heuristic algorithms optimize several well-known objective constraints, such as difficulty degree, number questions, etc., for producing optimal solutions. However, real scenarios, considering...
Different from the current node-level anomaly detection task, goal of graph-level is to find abnormal graphs that significantly differ others in a graph set. Due scarcity research on work detection, detailed description insufficient. Furthermore, existing works focus capturing anomalous information learn better representations, but they ignore importance an effective score function for evaluating graphs. Thus, this work, we first define including node and property anomalies set adopt...
In order to analyse the real-world information network more effectively, we propose a hierarchical embedding method based on partitioning, NPHNE. NPHNE is nested algorithm that can be combined with existing baseline algorithms enhance their representation. consists of two parts: graph abstracting and propagation. The process as follows: Firstly, modularity used pre-determine partition, purpose constrain maximum number levels. Then, hybrid collapsing method, series abstract graphs...