Unveiling the Core of Materials Properties via SISSO and Sensitivity Analysis

TL;DR

This paper enhances the interpretability of SISSO symbolic regression in materials science by introducing a sensitivity analysis that clarifies gene contributions and reveals key physical parameters influencing perovskite lattice constants.

Contribution

It introduces a derivative-based sensitivity analysis to resolve non-uniqueness in SISSO models, improving physical interpretability and insight into materials properties.

Findings

Identifies valence orbital radii, nuclear charges, and their products as key factors for perovskite lattice constants.

Demonstrates how different gene combinations encode equivalent physical information.

Enhances interpretability of SISSO models through sensitivity analysis.

Abstract

Interpretable AI can reveal physical principles governing intricate materials properties by uncovering explicit relationships between physical parameters and target properties. The sure-independence screening and sparsifying operator (SISSO) symbolic-regression approach identifies analytical expressions that correlate a target property with a small set of parameters, termed materials genes, selected from a large pool of candidates. However, multiple gene combinations can yield equally accurate SISSO models, with individual genes contributing with different weights. Here, we establish a derivative-based sensitivity analysis that resolves the non-uniqueness of symbolic-regression descriptions, enhances interpretability, thereby enabling deeper physical insight. This analysis reveals how distinct gene combinations encode equivalent information and identifies valence orbital radii, nuclear charges, and their products as the key quantities governing the equilibrium lattice constant of perovskites.