We are glad to present "E(n) Equivariant Graph Neural Networks". A new simple & effective method to build E(n) equivariance into your graphs. We will present it at ICML. Code is now also available at: github.com/vgsatorras/egnn. Joint work with @emiel_hoogeboom and @wellingmax.
@vgsatorras @emiel_hoogeboom @wellingmax Excuse me, dear professors, I’m a student from china and after reading your article I have a question about the Permutation equivariance. How did you prove it, I can’t find it in the article.
@Pressure_upon @vgsatorras @wellingmax Hi, we did not explicitly prove the permutation equivariance, since this is a feature that standard GNNs already have. You could derive it though, looking at Eqs. 3-6 and showing that a permutation of nodes in the previous layer, result in a permutation in the next layer
@emiel_hoogeboom @vgsatorras @wellingmax Thank you very much😁, it was very helpful for me!