Algorithm

Mathematical physics

Receptive field

Theoretical computer science

Scalability

Computer science

Acoustics

Physics

Database

Graph

Pattern recognition (psychology)

Mathematics

Invariant (physics)

Path (computing)

Artificial intelligence

Programming language

Artificial neural network

Permutation (music)

Proceedings of the ... AAAI Conference on Artificial Intelligence

We present, GeniePath, a scalable approach for learning adaptive receptive fields of neural networks defined on permutation invariant graph data. In GeniePath, we propose an adaptive path layer consists of two complementary functions designed for breadth and depth exploration respectively, where the former learns the importance of different sized neighborhoods, while the latter extracts and filters signals aggregated from neighbors of different hops away. Our method works in both transductive and inductive settings, and extensive experiments compared with competitive methods show that our approaches yield state-of-the-art results on large graphs.

GeniePath: Graph Neural Networks with Adaptive Receptive Paths