ASGN: An Active Semi-supervised Graph Neural Network for Molecular Property Prediction.
Zhongkai HaoChengqiang LuZhenya HuangHao WangZheyuan HuQi LiuEnhong ChenCheekong LeePublished in: KDD (2020)
Keyphrases
- neural network
- semi supervised
- prediction model
- protein function prediction
- semi supervised classification
- label propagation
- semi supervised learning
- graph construction
- prediction accuracy
- labeled data
- graph properties
- multi layer perceptron
- random walk
- graph structure
- unlabeled data
- graph theory
- directed graph
- global consistency
- graph representation
- three dimensional
- directed acyclic graph
- recurrent neural networks
- artificial neural networks
- graph theoretic
- back propagation
- supervised learning
- elman network
- pairwise constraints
- protein structure prediction
- drug design
- neural network ensemble
- manifold learning
- neural network model
- subspace learning
- graph mining
- protein structure
- bp neural network
- structured data
- fuzzy logic
- knn
- active learning
- chaotic time series
- anti monotonic
- pairwise