Recommended article: Evolution of artificial intelligence for application in contemporary materials science
In our webinars, we have often discussed applications of machine learning and deep learning to CT image segmentation analysis. In recent years, these types of artificial intelligence (AI) have advanced at a pace we've never seen before. Its applications seem ubiquitous.
ChatGPT is not writing this newsletter, but I used Gemini, Google's AI tool, to search for interesting research articles and found this review on the recent advancement of artificial intelligence for application in materials science.
The two main areas of AI application for contemporary materials science are forward modeling for predictive analysis and inverse modeling for optimization and design. The former helps accelerate the materials discovery process by generating new models with a higher probability of showing desired properties and simulating their performance before physically producing new materials. The latter helps us understand the underlying correlations between a large amount of data and use the gained insight to optimize process and materials design.
The paper reviews the evolution of AI, starting from traditional machine learning, conventional deep learning, and graph neural networks. Each section has an excellent summary of each algorithm and descriptions of their suitability. If you are thinking about using AI for your materials research but are not sure which network to use, this article might be a good place to start.
"Evolution of artificial intelligence for application in contemporary materials science," Vishu Gupta, Wei-keng Liao, Alok Choudhary, and Ankit Agrawal, MRS Communications 13, 754–763 (2023), https://doi.org/10.1557/s43579-023-00433-3
👓 Read the article
