vthost / DAGNN
A graph neural network tailored to directed acyclic graphs that outperforms conventional GNNs by leveraging the partial order as strong inductive bias besides other suitable architectural features.
☆118Updated 6 months ago
Related projects ⓘ
Alternatives and complementary repositories for DAGNN
- [NeurIPS 2023] Implementation of "Transformers over Directed Acyclic Graphs"☆54Updated 4 months ago
- Representing Long-Range Context for Graph Neural Networks with Global Attention☆123Updated 2 years ago
- Parameterized Explainer for Graph Neural Network☆128Updated 8 months ago
- ☆54Updated 3 years ago
- NAGphormer: A Tokenized Graph Transformer for Node Classification in Large Graphs☆112Updated 8 months ago
- PyTorch code of "SLAPS: Self-Supervision Improves Structure Learning for Graph Neural Networks"☆85Updated 3 years ago
- A graph transformer framework☆75Updated 2 years ago
- The source code for NeurIPS 2020 paper "Graph Policy Network for Transferable Active Learning on Graphs"☆44Updated 4 years ago
- Official Code Repository for the paper "Accurate Learning of Graph Representations with Graph Multiset Pooling" (ICLR 2021)☆101Updated 2 years ago
- Code of "Breaking the Limits of Message Passing Graph Neural Networks" paper published in ICML2021☆40Updated 3 years ago
- Official code of "Towards Multi-Grained Explainability for Graph Neural Networks" (NeurIPS 2021) + Pytorch Implementation of recent attri…☆67Updated last year
- Graph Posterior Network: Bayesian Predictive Uncertainty for Node Classification (NeurIPS 2021)☆40Updated 2 years ago
- Boost learning for GNNs from the graph structure under challenging heterophily settings. (NeurIPS'20)☆97Updated 2 years ago
- AAAI'21: Data Augmentation for Graph Neural Networks