Théo Vincent

Théo Vincent focuses on developing autonomous learning algorithms that enable agents to acquire relevant skills in any environment through direct interaction, starting without prior knowledge. These algorithms enhance traditional temporal-difference methods by accelerating skill acquisition (i-QN, and PBO), reducing dependency on hyperparameters (AdaQN), and lowering resource requirements (EauDeQN, and iS-QN). The goal is to create more efficient, adaptable, and practical reinforcement learning systems for real-world applications.

Théo joined the Intelligent Autonomous Systems Group as a Ph.D. student in December 2022. He is part of the group SAIROL, a DFKI unit, and works under the supervision of Jan Peters. During his studies, Théo focused on computer vision and reinforcement learning. He graduated from MVA, a double master's degree program in collaboration with ENS ParisSaclay and Ponts ParisTech. He did his master's thesis at IAS on value-based methods in offline settings supervised by Carlo D’Eramo.

Before his Ph.D., Théo worked in a Parisian lab called Saint-Venant lab, supervised by Rémi Carmigniani to track professional swimmers on video clips using Deep Learning. He then joined Signality, a Swedish start-up led by Mikael Rousson, to investigate the problem of finding a homography linking a football pitch to a camera. He also worked in the group of Chirag Patel at Harvard Medical School to understand how our organs are aging.

Publications

1. •     Bib
     Vincent, T.; Tripathi, Y.; Faust, T.; Akgül, A.; Oren, Y.; Kandemir, M.; Peters, J.; D'Eramo, C. (2026). Bridging the Performance Gap Between Target-Free and Target-Based Reinforcement Learning, International Conference on Learning Representations (ICLR).
2. •     Bib
     Hendawy, A.; Metternich, H.; Vincent, T.; Kallel, M.; Peters, J.; D'Eramo, C. (2026). Use the Online Network If You Can: Towards Fast and Stable Reinforcement Learning, International Conference on Learning Representations (ICLR).
3. •     Bib
     Vincent, T.; Faust, T.; Tripathi, Y.; Peters, J.; D'Eramo, C. (2025). Eau De Q-Network: Adaptive Distillation of Neural Networks in Deep Reinforcement Learning, Reinforcement Learning Journal (RLJ).
4. •     Bib
     Vincent, T.; Palenicek, D.; Belousov, B.; Peters, J.; D'Eramo, C. (2025). Iterated Q-Network: Beyond One-Step Bellman Updates in Deep Reinforcement Learning, Transactions on Machine Learning Research (TMLR), J2C Certificate.
5. •     Bib
     Vincent, T.; Wahren, F.; Peters, J.; Belousov, B.; D'Eramo, C. (2025). Adaptive Q-Network: On-the-fly Target Selection for Deep Reinforcement Learning, International Conference on Learning Representations (ICLR).
6. •     Bib
     Vincent, T.; Metelli, A.; Belousov, B.; Peters, J.; Restelli, M.; D'Eramo, C. (2024). Parameterized Projected Bellman Operator, Proceedings of the National Conference on Artificial Intelligence (AAAI).

Workshops organization

1. Döring L.; Vincent, T.; Weißmann, S.; Vernade, C. (2026). Reinforcement Learning in Mannheim.