At the 19th International Symposium on Novel and Nano Materials (ISNNM 2026), held in Dresden, Germany, from July 12 to 17, Lee Woo-young, a combined master’s and Ph.D. student in the Department of New Materials Engineering at Kookmin University (advisor: Professor Choi HyunJoo) from the Nano-Convergence Structural Materials Laboratory, received the ASFLOW Best Presentation Award, showcasing his outstanding research achievements.

7000-series (Al-Zn-Mg-Cu) aluminum alloys are used as key structural materials in the aerospace and transportation sectors due to their excellent strength-to-weight ratio, and their mechanical properties are determined by the microstructure of nano-precipitates formed during the aging heat treatment process. However, since precipitation behavior varies complexly depending on the alloy composition and heat treatment conditions, finding the optimal composition-process combination has traditionally required extensive experimental trial and error. Recently, data-driven materials design utilizing machine learning has garnered attention as an alternative to overcome these limitations; however, the significant time and cost required to obtain reliable experimental data remain major constraints. Consequently, active learning strategies—which enable efficient training with limited data and allow the model to autonomously propose future experiments—are emerging as an important research topic.

Accordingly, Lee Woo-Young, a Ph.D. candidate in the integrated master’s-Ph.D. program, delivered an oral presentation titled “Rapid Prediction of Precipitate Microstructures in 7000-Series Aluminum Alloys Using DSC–TEM Correlation and Gaussian Process Regression.” This study combines features based on CALPHAD thermodynamic calculations with a Gaussian process regression (GPR) model to predict the precipitate microstructure of 7xxx-series aluminum alloys, and It is particularly significant because the study demonstrated that features incorporating thermodynamic information significantly improve the data exploration efficiency of active learning while maintaining prediction accuracy. By doing so, it presents a methodology that can drastically reduce the number of experiments required for alloy design and heat treatment process optimization—even in environments with limited experimental data—thereby accelerating the practical application of data-driven alloy design.