Timezone: »

Graph Neural Networks with Adaptive Readouts
David Buterez · Jon Paul Janet · Steven J Kiddle · Dino Oglic · Pietro Liò

Tue Nov 29 02:00 PM -- 04:00 PM (PST) @ Hall J #321

An effective aggregation of node features into a graph-level representation via readout functions is an essential step in numerous learning tasks involving graph neural networks. Typically, readouts are simple and non-adaptive functions designed such that the resulting hypothesis space is permutation invariant. Prior work on deep sets indicates that such readouts might require complex node embeddings that can be difficult to learn via standard neighborhood aggregation schemes. Motivated by this, we investigate the potential of adaptive readouts given by neural networks that do not necessarily give rise to permutation invariant hypothesis spaces. We argue that in some problems such as binding affinity prediction where molecules are typically presented in a canonical form it might be possible to relax the constraints on permutation invariance of the hypothesis space and learn a more effective model of the affinity by employing an adaptive readout function. Our empirical results demonstrate the effectiveness of neural readouts on more than 40 datasets spanning different domains and graph characteristics. Moreover, we observe a consistent improvement over standard readouts (i.e., sum, max, and mean) relative to the number of neighborhood aggregation iterations and different convolutional operators.

Author Information

David Buterez (University of Cambridge)
Jon Paul Janet (AstraZeneca)
Steven J Kiddle (AstraZeneca)

Director, Health Data Science in the Healthcare Analytics team, in the Data Science & Advanced Analytics department. Lead the real world insights/evidence activities for respiratory & immunology therapy area in R&D Biopharmaceuticals AstraZeneca.

Dino Oglic (AstraZeneca)
Pietro Liò (University of Cambridge)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors