Graph Neural Networks
Tim and Zach delve into the world of Graph Neural Networks, drawing parallels to cellular automata and discussing the continuous versus discrete nature of information propagation. They explore the convergence challenges in training models and the potential benefits of breaking the isotropic diffusion assumption in GNNs based on insights from Bronstein's work on partial differential equations.In this clip
From this podcast
Machine Learning Street Talk (MLST)
#71 - ZAK JOST (Graph Neural Networks + Geometric DL) [UNPLUGGED]
Related Questions
Can formal systems be computed as discussed in the episode Stephen Wolfram: Cellular Automata, Computation, and Physics?
Can formal systems be computed as discussed in the episode Stephen Wolfram: Cellular Automata, Computation, and Physics | Lex Fridman Podcast #89 and the clip Computational Foundations of Physics?