Towards Interpretable Sparse Graph Representation Learning with Laplacian Pooling
This addresses interpretability and efficiency issues in molecular representation learning for drug design, though it appears incremental as it builds on existing GNN pooling methods.
The authors tackled the problem of graph neural networks (GNNs) failing to capture the importance of interactions between molecular substructures by proposing LaPool, a hierarchical graph pooling method, which outperformed recent GNNs on molecular graph prediction tasks and remained competitive on non-molecular tasks.
Recent work in graph neural networks (GNNs) has led to improvements in molecular activity and property prediction tasks. Unfortunately, GNNs often fail to capture the relative importance of interactions between molecular substructures, in part due to the absence of efficient intermediate pooling steps. To address these issues, we propose LaPool (Laplacian Pooling), a novel, data-driven, and interpretable hierarchical graph pooling method that takes into account both node features and graph structure to improve molecular representation. We benchmark LaPool on molecular graph prediction and understanding tasks and show that it outperforms recent GNNs. Interestingly, LaPool also remains competitive on non-molecular tasks. Both quantitative and qualitative assessments are done to demonstrate LaPool's improved interpretability and highlight its potential benefits in drug design. Finally, we demonstrate LaPool's utility for the generation of valid and novel molecules by incorporating it into an adversarial autoencoder.