This was part of The Multifaceted Complexity of Machine Learning

Learning in Graph Neural Networks

Stefanie Jegelka, Massachussetts Institute of Technology

Thursday, April 15, 2021

Abstract: Graph Neural Networks (GNNs) have become a popular tool for learning representations of graph-structured inputs, with applications in computational chemistry, recommendation, pharmacy, reasoning, and many other areas. In this talk, I will show some recent results on learning with message-passing GNNs. In particular, GNNs possess important invariances and inductive biases that affect learning and generalization. We relate these properties and the choice of the “aggregation function” to predictions within and outside the training distribution.