Pireci Sejdiu, Nora and Rendevski, Nikola and Ristevski, Blagoj (2025) AI-Based Prediction of Elastic Properties in Crystals with Class Balancing. In: International Conference on Applied Internet and Information Technologies (AIIT 2025), 7.11.2015, Bitola, Republic of Macedonia.
![[thumbnail of AIIT 2025_NPS_NR_BR.pdf]](https://eprints.uklo.edu.mk/style/images/fileicons/text.png) |
Text
AIIT 2025_NPS_NR_BR.pdf Download (1MB) |
Abstract
For the design of new functional materials for industrial applications, predicting the mechanical properties of materials is essential. In this study, we present an AI-powered classification framework for predicting the elastic moduli of crystalline materials based on the elastic_tensor_2015 dataset provided by theMaterials Project, a publicly available repository that provides computed properties of a wide range of crystalline materials. The biggest challenge of this dataset is class imbalance, which can bias the predictive models and limit generalisation. To address this issue, we applied two oversampling techniques - ADASYN and SMOTE-Tomek - to generate synthetic minority samples and improve model learning. We trained and evaluated eight machine learning algorithms, including Balanced Random Forest, XGBoost, CatBoost, Multi-Layer Perceptron (MLP), Logistic Regression, K-Nearest Neighbors, Support Vector Machine, and Naive Bayes. Model evaluation was performed using standard metrics such as precision, recall, F1 score, and ROC-AUC, along with feature importance analysis to interpret model decisions. The results show that ensemble-based models and neural networks achieved the highest predictive performance, while simpler models such as SVM and Naive Bayes showed limited effectiveness. This study highlights the impact of data balancing and algorithm selection on machine learning models for materials informatics and provides insights into the development of accurate, data-driven tools to accelerate material discovery.
| Item Type: | Conference or Workshop Item (Paper) |
|---|---|
| Subjects: | Scientific Fields (Frascati) > Natural sciences > Computer and information sciences Scientific Fields (Frascati) > Engineering and Technology > Electrical engineering, electronic engineering,information engineering |
| Divisions: | Faculty of Information and Communication Technologies |
| Depositing User: | Prof. d-r. Blagoj Ristevski |
| Date Deposited: | 10 Dec 2025 10:22 |
| Last Modified: | 10 Dec 2025 10:22 |
| URI: | https://eprints.uklo.edu.mk/id/eprint/11266 |
Actions (login required)
 |
View Item |