DESCRIPTION
The application of artificial intelligence (AI) and machine learning (ML) techniques in materials science is a rapidly evolving field, with significant potential for accelerating materials discovery and design. Recently a shift towards automated, AI-driven processes in materials science, is shaped owing to the need for accelerated scientific discovery cycles. V. Gupta et al [[1]] highlight the transformative impact of AI in this domain, particularly in forward modeling for predictive analysis and inverse modeling for optimization and design. H. J. Kulik et al [[2]] further underscores the critical progress in AI and ML approaches for various aspects of computational materials science, including electron microscopy, energy materials design, and crystal nucleation and growth. L. Goswami et al [[3]] provides a comprehensive analysis of AI’s role in material engineering, emphasizing its potential in property prediction, material processing, and performance analysis. A. Merchan et al [[4]] shows that the use of graph neural networks trained at scale, improves materials discovery efficiency, in combination with ab initio calculations and iteratively filtering candidate structures through DFT verification. Diverse candidate structures were generated using symmetry-aware partial substitutions (SAPS) and random structure search. The energy of filtered candidates was computed using DFT for model validation and training data generation. A significant expansion is demonstrated in stable materials known to humanity, as well as the improvement in prediction accuracy and hit rate of stable predictions by graph networks for materials exploration (GNoME) models. Moreover E. O. Pyzer-Knapp et al [[5]], scrutinizes how new tools in materials science, driven by AI, simulation, and experimental automation, are transforming traditional manual processes into automated, parallel, and iterative processes to accelerate and enrich each stage of the discovery cycle. The phases of a typical materials discovery effort are highlighted, emphasizing the importance of specifying research questions, collecting relevant data, forming hypotheses, and conducting experiments to generate knowledge and create new hypotheses. These studies collectively underscore the growing importance of AI in materials science and its potential to revolutionize the field. Special focus is placed on Machine Learning-Based Interatomic Potentials[[6]].
[[1]] Gupta, V., Liao, Wk., Choudhary, A. et al. Evolution of artificial intelligence for application in contemporary materials science. MRS Communications 13, 754–763 (2023). https://doi.org/10.1557/s43579-023-00433-3
[[2]] Kulik, H.J., Tiwary, P. Artificial intelligence in computational materials science. MRS Bulletin 47, 927–929 (2022). https://doi.org/10.1557/s43577-022-00431-1
[[3]] Goswami, L., Deka, M.K., Roy, M. Artificial Intelligence in Material Engineering: A Review on Applications of Artificial Intelligence in Material Engineering. Advanced Engineering Materials, 25(13), 2300104 (2023). https://doi.org/10.1002/adem.202300104
[[4]] Merchant, A., Batzner, S., Schoenholz, S.S., et al Scaling deep learning for materials discovery. Nature, 624(7990), pp. 80–85 (2023) https://doi.org/10.1038/s41586-023-06735-9
[[5]] Pyzer-Knapp, E.O., Pitera, J.W., Staar, P.W.J. et al Accelerating materials discovery using artificial intelligence, high performance computing and robotics. npj Computational Materials, 8(1), 84 (2022). https://doi.org/10.1038/s41524-022-00765-z
[[6]] E Nikidis, N Kyriakopoulos, R Tohid, K Kachrimanis, J Kioseoglou, Harnessing machine learning for efficient large-scale interatomic potential for sildenafil and pharmaceuticals containing H, C, N, O, and S, Nanoscale 16 (38), 18014-18026(2024) DOI:10.1039/D4NR00929K
DETAILS
Course type: Special Session (in person delivery)
Duration: 2 hours
Institution of lecturer: School of Physics, Department of Condensed Matter and Materials Physics, Aristotle University and Center for Interdisciplinary Research and Innovation, Aristotle University of Thessaloniki, Thessaloniki, Greece.
LECTURER
Joseph Kioseoglou has conducted research in Greece and abroad (Barcelona, Caen, Dusseldorf, Nancy, Grenoble, Okinawa, Lyon). He is interested in materials analysis and design by the use of atomistic simulations and advanced computational techniques and algorithms employed in materials science (Artificial intelligence, Machine learning, Deep learning, etc.). He has studied several semiconducting and metallic materials in crystalline and amorphous phases and he has investigated extended defects, surfaces, interfaces and nanostructures (nanoparticles, quantum dots and wires) at the atomic scale, as well as he explored structure-property relationships in materials, focusing in generation of novel insights and applications. His research mainly focuses on structural and electronic properties, energetic stability, growth kinetics and surface thermodynamics and he has actively participated in research projects that involved developing and optimizing computational models to predict material properties. He has participated in more than 20 EU and national research projects.