A5083903843- Advanced Graph Neural Networks
- Bayesian Modeling and Causal Inference
- Topic Modeling
- Recommender Systems and Techniques
- Sentiment Analysis and Opinion Mining
- Graph theory and applications
- Matrix Theory and Algorithms
- Virus-based gene therapy research
- Herpesvirus Infections and Treatments
- Stochastic Gradient Optimization Techniques
- Rough Sets and Fuzzy Logic
- Complex Network Analysis Techniques
- Animal Virus Infections Studies
- Machine Learning and Algorithms
- Spectral Theory in Mathematical Physics
- Machine Learning and Data Classification
Renmin University of China
2021-2024
Guangxi Center for Disease Prevention and Control
2015
Guangxi University
2015
China Animal Disease Control Center
2015
Academy of Military Medical Sciences
2015
Many representative graph neural networks, e.g., GPR-GNN and ChebNet, approximate convolutions with spectral filters. However, existing work either applies predefined filter weights or learns them without necessary constraints, which may lead to oversimplified ill-posed To overcome these issues, we propose BernNet, a novel network theoretical support that provides simple but effective scheme for designing learning arbitrary In particular, any over the normalized Laplacian spectrum of graph,...
Designing spectral convolutional networks is a challenging problem in graph learning. ChebNet, one of the early attempts, approximates convolutions using Chebyshev polynomials. GCN simplifies ChebNet by utilizing only first two polynomials while still outperforming it on real-world datasets. GPR-GNN and BernNet demonstrate that Monomial Bernstein bases also outperform basis terms learning convolutions. Such conclusions are counter-intuitive field approximation theory, where established...
Efficient computation of node proximity queries such as transition probabilities, Personalized PageRank, and Katz are fundamental importance in various graph mining learning tasks. In particular, several recent works leverage fast to improve the scalability Graph Neural Networks (GNN). However, prior studies on GNN feature propagation a case-by-case basis, with each paper focusing particular measure. this paper, we propose Approximate Propagation (AGP), unified randomized algorithm that...
Spectral Graph Neural Networks have demonstrated superior performance in graph representation learning. However, many current methods focus on employing shared polynomial coefficients for all nodes, i.e., learning node-unified filters, which limits the filters' flexibility node-level tasks. The recent DSF attempts to overcome this limitation by node-wise based positional encoding. initialization and updating process of encoding are burdensome, hindering scalability large-scale graphs. In...
Heterogeneous Graph Neural Networks (HGNNs) have gained significant popularity in various heterogeneous graph learning tasks. However, most HGNNs rely on spatial domain-based message passing and attention modules for information propagation aggregation. These spatial-based neglect the utilization of spectral convolutions, which are foundation Convolutional (GCN) homogeneous graphs. Inspired by effectiveness scalability spectral-based GNNs graphs, this paper explores extension to We propose...