A brief overview of deep generative models and how they can be used to discover new electrode materials

Abstract

As humankind searches for sustainable energy solutions, the demand for electrochemistry has increased. Thus, new and more advanced electrode materials are required. However, finding electrodes that meet the necessary performance is a challenge. Machine learning models can predict key properties such as catalytic activity and stability with surprisingly good accuracy, thus accelerating the process of evaluating materials. However, in most cases, the same models cannot explain how to generate new material compositions. Here, deep generative models can become very valuable. Although issues related to data availability and understanding how these models work still exist, combining deep generative models with computer simulations and laboratory experiments hold great potential for developing the next generation of electrodes. This short review will show recent progress in using deep generative models in related material fields and stress how these models can accelerate the discovery of electrode materials.