Machine learning approach to topological graph descriptors of graphene nanoribbons

TL;DR

This paper uses machine learning to analyze the structural features of graphene nanoribbons and predict their properties for nanotechnology applications.

Contribution

The study introduces a machine learning framework using valency-based molecular descriptors to predict graphene nanoribbon properties.

Findings

Valency-based molecular descriptors were computed and evaluated for predictive power.

Logistic regression with ROC analysis showed the effectiveness of these descriptors in capturing structural features.

The framework helps design graphene nanoribbons with desired functionalities.

Abstract

Bottom-up syntheses of graphene nanoribbons have gathered considerable research interest because of their electronic properties and quantum behaviors which enhance their significance in nanotechnology. The advancement of material’s design methods and applications heavily depends on understanding how structural topology influences functional properties. This study analyzed valency based molecular descriptors of graphene nanoribbons with machine learning techniques. We have computed various valency based molecular descriptors of graphene nanoribbons and studied their predictive power using the logistic regression machine learning technique by describing its receiver operating characteristic curve to analyze the topological features of these graphene nanoribbons. These descriptors represent quantitative measurements of crucial structural features of graphene nanoribbons that directly affect material properties. This predictive framework enables researchers to design graphene nanoribbons with specific functionalities while advancing their knowledge about structure-property relationships in this material class. Molecular descriptors combined with machine learning methods demonstrate the potential to accelerate the discovery process and optimization of advanced nanomaterials.

Figures