About Me
Hello! I’m Dimitri, a third-year PhD student in Computer Science at Johns Hopkins University (JHU), advised by Prof. Nicolas Loizou. I’m broadly interested in large-scale optimization and machine learning. My current research focuses on adaptive methods for minimization problems, with applications in deep learning and large language models.
Before JHU, I earned a BSc in Mathematics from the National and Kapodistrian University of Athens (NKUA), an MSc in Data Science and Machine Learning from the National Technical University of Athens (NTUA), and an MSc in Mathematics from Göttingen University.
Research
Safeguarded Stochastic Polyak Step-Sizes for Non-smooth Optimization: Robust Performance Without Small (Sub)Gradients
Under Submission
Dimitris Oikonomou, Nicolas Loizou
Analysis of an idealized stochastic polyak method and its application to black-box model distillation
In arXiv:2504.01898, 2025
Robert Gower, Guillaume Garrigos, Nicolas Loizou, Dimitris Oikonomou, Konstantin Mishchenko, Fabian Schaipp
Sharpness-Aware Minimization: General Analysis and Improved Rates
In the 13th International Conference on Learning Representations (ICLR 2025), Code
Dimitris Oikonomou, Nicolas Loizou
Stochastic Polyak Step-sizes and Momentum: Convergence Guarantees and Practical Performance
In the 13th International Conference on Learning Representations (ICLR 2025), Code
Dimitris Oikonomou, Nicolas Loizou
Selected Honors & Awards
Mathematical Olympiads
- International Mathematical Olympiad (IMO): Bronze Medal @ Santa Marta, Colombia, 2013
- Greek Mathematical Olympiad: Silver Medal @ Athens, Greece, 2014
- Greek Mathematical Olympiad: Bronze Medal @ Athens, Greece, 2013
Scholarships
- Hellenic Post Scholarship: 2015-2019
- DAAD Scholarship: 2019-2020
- JHU MINDS Fellowship: 2024, 2025
Invited Talks & Posters
- International Conference on Machine Learning (ICML 2025), Singapore (2x Poster)
- SIAM Conference on Optimization 2026, Edinburgh, UK
Teaching & Professional Service
- Teaching Assistant: Computer Vision @ JHU, Fall 2024
- Reviewer: IEEE Transactions on Signal Processing, ICML, NeurIPS, ICLR