Self-supervised Training of Graph Convolutional Networks
動機： GCNs需要鄰接矩陣作爲輸入來定義這些非網格數據之間的關係，這就導致所有的數據，包括訓練數據、驗證數據和測試數據，通常只形成一個用於訓練的圖結構數據。此外，鄰接矩陣通常是預定義的且平穩的，這使得數據增強策略不能在構造的圖結構數據上使用來增加訓練數據的數量。爲了進一步提高有限訓練數據下的學習能力和模型性能 貢獻： 提出從輸入圖結構數據本身中挖掘可用信息，以提高基於圖的模型的學習識別能力和性能
相關文章
- 1. Modification of Graph Convolutional Networks in PyTorch
- 2. HyperGCN: A New Method of Training Graph Convolutional Networks on Hypergraphs--論文筆記
- 3. Applications of Graph Neural Networks
- 4. MuchGCN：Multi-Channel Graph Convolutional Networks
- 5. Emotion Recognition Using Graph Convolutional Networks
- 6. Adaptive Graph Convolutional Neural Networks
- 7. 《Large-Scale Learnable Graph Convolutional Networks》
- 8. Semi-Supervised Classification With Graph Convolutional Networks
- 9. STRATEGIES FOR PRE-TRAINING GRAPH NEURAL NETWORKS
- 10. Strategies For Pre-Training Graph Neural Networks
- 更多相關文章...
- • XSLT
- • XSLT
- • JDK13 GA發佈：5大特性解讀
- • 使用Rxjava計算圓周率
相關標籤/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. Modification of Graph Convolutional Networks in PyTorch
- 2. HyperGCN: A New Method of Training Graph Convolutional Networks on Hypergraphs--論文筆記
- 3. Applications of Graph Neural Networks
- 4. MuchGCN：Multi-Channel Graph Convolutional Networks
- 5. Emotion Recognition Using Graph Convolutional Networks
- 6. Adaptive Graph Convolutional Neural Networks
- 7. 《Large-Scale Learnable Graph Convolutional Networks》
- 8. Semi-Supervised Classification With Graph Convolutional Networks
- 9. STRATEGIES FOR PRE-TRAINING GRAPH NEURAL NETWORKS
- 10. Strategies For Pre-Training Graph Neural Networks