The NODA – Numerical Optimization and Data Analysis Research Group conducts research at the intersection of numerical optimization, numerical linear algebra, and data analysis, with a strong emphasis on the development of theoretically grounded algorithms for large-scale and data-driven problems. The group’s work spans deterministic, stochastic, and randomized optimization methods, including trust-region schemes, sketching techniques, adaptive second-order stochastic algorithms, and objective-function-free approaches, with rigorous convergence and complexity analysis.

Recent contributions address optimization methods tailored to machine learning and data science, including algorithms for training neural networks, multiclass data segmentation, sparse support vector machines, and multi-task learning.

The NODA group combines solid mathematical foundations with extensive numerical experimentation, often releasing accompanying open-source software to ensure reproducibility and practical impact. Through its publications in international journals, preprint archives, and optimization platforms, the group contributes to advancing scalable and reliable computational methods for modern optimization and data analysis challenges.