Moderate Message Passing Improves Calibration: A Universal Way to Mitigate Confidence Bias in Graph Neural Networks.
Min WangHao YangJincai HuangQing ChengPublished in: AAAI (2024)
Keyphrases
- message passing
- neural network
- belief propagation
- distributed systems
- factor graphs
- approximate inference
- junction tree
- probabilistic inference
- inference in graphical models
- tree reweighted
- distributed shared memory
- shared memory
- singly connected
- max product
- sum product algorithm
- sum product
- graphical models
- matrix multiplication
- markov random field
- particle filter
- ldpc codes
- computer vision