Energy-GNoME Featured in Energy and AI

We are pleased to announce that our work "Energy-GNoME: A living database of selected materials for energy applications" has been published in the peer-reviewed journal Energy and AI.

This publication marks a key milestone in the development of Energy-GNoME, presenting the scientific foundation and methodology behind our open, AI-driven database for accelerated discovery of materials for energy applications.

About the Paper

In the article, authored by Paolo De Angelis, Giulio Barletta, Giovanni Trezza, Pietro Asinari, and Eliodoro Chiavazzo, we introduce an artificial-intelligence protocol that screens vast, unexplored materials spaces — such as the GNoME database — to identify candidates for thermoelectrics, perovskite photovoltaics, and battery cathodes.

The workflow combines:

• Binary classifiers (“AI-experts”) that mitigate cross-domain bias and improve reliability.
• Regressors that predict key material properties such as the thermoelectric figure of merit (\(zT\)), band gap (\(E_g\)), and cathode voltage (\(\Delta V_c\)).

By applying this protocol to the GNoME dataset, we identified over 38,500 new candidate materials for energy applications — accessible through the Energy-GNoME dashboards.

A Living, Evolving Database

Energy-GNoME is designed as a living database: its models and datasets are meant to be refined iteratively through community validation and feedback. Researchers can explore candidate materials, contribute new data, and help improve the reliability and scope of AI-based materials discovery.

The publication emphasizes this collaborative approach, inviting both computational and experimental scientists to engage in the continuous “active learning” cycle that drives Energy-GNoME's evolution.