Neural Network-Guided Symbolic Regression for Interpretable Descriptor Discovery in Perovskite Catalysts

TL;DR

This paper introduces a neural network-guided symbolic regression framework that effectively discovers interpretable and accurate descriptors for oxide perovskite catalysts' activity, even with limited data, enhancing materials informatics.

Contribution

The study presents a novel two-phase approach combining neural networks and symbolic regression to identify physically meaningful descriptors in small datasets for catalyst activity prediction.

Findings

Reproduced and improved the {bc}/t descriptor with small dataset.

Identified LUMO energy as a key electronic descriptor.

Achieved accurate predictions with physically interpretable formulas.

Abstract

Understanding and predicting the activity of oxide perovskite catalysts for the oxygen evolution reaction (OER) requires descriptors that are both accurate and physically interpretable. While symbolic regression (SR) offers a path to discover such formulas, its performance degrades with high-dimensional inputs and small datasets. We present a two-phase framework that combines neural networks (NN), feature importance analysis, and symbolic regression (SR) to discover interpretable descriptors for OER activity in oxide perovskites. In Phase I, using a small dataset and seven structural features, we reproduce and improve the known {\mu}/t descriptor by engineering composite features and applying symbolic regression, achieving training and validation MAEs of 22.8 and 20.8 meV, respectively. In Phase II, we expand to 164 features, reduce dimensionality, and identify LUMO energy as a key electronic descriptor. A final formula using {\mu}/t, {\mu}/RA, and LUMO energy achieves improved accuracy (training and validation MAEs of 22.1 and 20.6 meV) with strong physical interpretability. Our results demonstrate that NN-guided symbolic regression enables accurate, interpretable, and physically meaningful descriptor discovery in data-scarce regimes, indicating interpretability need not sacrifice accuracy for materials informatics.