Graph neural networks and MSO
By: Veeti Ahvonen, Damian Heiman, Antti Kuusisto
Potential Business Impact:
Lets computers understand patterns in tree-like data.
We give an alternative proof for the existing result that recurrent graph neural networks working with reals have the same expressive power in restriction to monadic second-order logic MSO as the graded modal substitution calculus. The proof is based on constructing distributed automata that capture all MSO-definable node properties over trees. We also consider some variants of the acceptance conditions.
Similar Papers
On Expansions of Monadic Second-Order Logic with Dynamical Predicates
Logic in Computer Science
Lets computers check number patterns automatically.
A finer reparameterisation theorem for MSO and FO queries on strings
Logic in Computer Science
Finds patterns in text faster.
The Correspondence Between Bounded Graph Neural Networks and Fragments of First-Order Logic
Artificial Intelligence
Lets computers understand complex connections better.