Calibrate and Debias Layer-wise Sampling for Graph Convolutional Networks

Keyphrases

• Algorithm Implementation 50%
• Code Implementation 50%
• Debiasing 100%
• Embedding Aggregation 50%
• Estimation Bias 50%
• Estimation Experiment 50%
• Graph Convolutional Network 100%
• Importance Sampling 50%
• Layerwise Theory 50%
• Matrix Approximation 50%
• Multiple Samples 50%
• Network Training 50%
• New Principle 50%
• Node Embedding 50%
• Probability Bias 50%
• Publicly Available 50%
• Sample Estimation 50%
• Sampling Methods 100%
• Sampling Probability 100%
• Sampling without Replacement 50%
• Variance Estimation 50%

Mathematics

• Approximated Matrix 50%
• Graph Convolutional Network 100%
• Importance Sampling 50%
• Probability Theory 100%
• Sampling Without Replacement 50%
• Variance Estimation 50%

Computer Science

• Approximated Matrix 50%
• Estimation Variance 50%
• Graph Convolutional Network 100%
• Importance Sampling 50%
• Node Embedding 50%