Multi-View Polymer Representations for the Open Polymer Prediction

TL;DR

This paper introduces a multi-view approach combining various polymer representations, including descriptors, graph neural networks, 3D data, and language models, to improve property prediction accuracy in a large-scale challenge.

Contribution

It presents a novel ensemble method that integrates four diverse polymer representations, achieving competitive results in the Open Polymer Prediction Challenge.

Findings

Achieved a top-10 ranking in the NeurIPS 2025 challenge.

Attained a public MAE of 0.057 and private MAE of 0.082.

Demonstrated the effectiveness of multi-view ensemble in polymer property prediction.

Abstract

We address polymer property prediction with a multi-view design that exploits complementary representations. Our system integrates four families: (i) tabular RDKit/Morgan descriptors, (ii) graph neural networks, (iii) 3D-informed representations, and (iv) pretrained SMILES language models, and averages per-property predictions via a uniform ensemble. Models are trained with 10-fold splits and evaluated with SMILES test-time augmentation. The approach ranks 9th of 2241 teams in the Open Polymer Prediction Challenge at NeurIPS 2025. The submitted ensemble achieves a public MAE of 0.057 and a private MAE of 0.082.