Showing 1–1 of 1 results for author: Rymarczyk, D

ProGReST: Prototypical Graph Regression Soft Trees for Molecular Property Prediction

Authors: Dawid Rymarczyk, Daniel Dobrowolski, Tomasz Danel

Abstract: In this work, we propose the novel Prototypical Graph Regression Self-explainable Trees (ProGReST) model, which combines prototype learning, soft decision trees, and Graph Neural Networks. In contrast to other works, our model can be used to address various challenging tasks, including compound property prediction. In ProGReST, the rationale is obtained along with prediction due to the model's bui… ▽ More In this work, we propose the novel Prototypical Graph Regression Self-explainable Trees (ProGReST) model, which combines prototype learning, soft decision trees, and Graph Neural Networks. In contrast to other works, our model can be used to address various challenging tasks, including compound property prediction. In ProGReST, the rationale is obtained along with prediction due to the model's built-in interpretability. Additionally, we introduce a new graph prototype projection to accelerate model training. Finally, we evaluate PRoGReST on a wide range of chemical datasets for molecular property prediction and perform in-depth analysis with chemical experts to evaluate obtained interpretations. Our method achieves competitive results against state-of-the-art methods. △ Less

Submitted 27 December, 2022; v1 submitted 7 October, 2022; originally announced October 2022.

Comments: Accepted to SDM2023