Razvan-Andrei Lascu

I am a Postdoctoral Researcher in the Deep Learning Theory team at RIKEN AIP. My research lies at the interface of Gradient Flows, Optimal Transport, Optimization and Reinforcement Learning.

I was previously a PhD student at Heriot-Watt University and University of Edinburgh, supervised by Mateusz Majka and Lukasz Szpruch.

Publications

R.-A. Lascu, M. B. Majka. Non-convex entropic mean-field optimization via Best Response flow. To appear in NeurIPS (2025).
R.-A. Lascu, M. B. Majka, Ł. Szpruch. Entropic mean-field min-max problems via Best Response flow. Appl. Math. Optim. 91, 48 (2025).
R.-A. Lascu, M. B. Majka, Ł. Szpruch. A Fisher-Rao gradient flow for entropic mean-field min-max games. Transactions on Machine Learning Research (2024).

Preprints

R.-A. Lascu, D. Šiška, Ł. Szpruch (2025). PPO in the Fisher-Rao geometry.
R.-A. Lascu, M. B. Majka, D. Šiška, Ł. Szpruch (2024). Linear convergence of proximal descent schemes on the Wasserstein space.
R.-A. Lascu, M. B. Majka, Ł. Szpruch (2024). Mirror Descent-Ascent for mean-field min-max problems.

Talks/Workshops/Conferences

Teaching

Contact