On Over-Squashing in Message Passing Neural Networks: The Impact of Width, Depth, and Topology.
Francesco Di GiovanniLorenzo GiustiFederico BarberoGiulia LuisePietro Lio'Michael M. BronsteinPublished in: CoRR (2023)
Keyphrases
- message passing
- neural network
- belief propagation
- distributed systems
- probabilistic inference
- shared memory
- inference in graphical models
- sum product algorithm
- factor graphs
- markov random field
- approximate inference
- interconnection networks
- graphical models
- depth map
- artificial neural networks
- stereo matching
- loopy belief propagation
- distributed shared memory
- graph cuts
- junction tree
- matrix multiplication
- ldpc codes
- pairwise
- sum product
- three dimensional