Anish Diwan

Anish Diwan joined the Intelligent Autonomous Systems Group as a Ph.D. student in April 2025. Currently, he works on imitation learning, inverse reinforcement learning, and reinforcement learning, with a focus on efficient, geometry-aware policy/reward updates and techniques like score-based generative models.

Before his PhD, Anish Diwan completed his Master's degree in Robotics at the Technische Universiteit Delft with Cum Laude honours. His thesis entitled “Noise-conditioned Energy-based Annealed Rewards" was written under the supervision of Prof. Jens Kober , Prof. Jan Peters, and Julen Urain.

Key References

1. •       Bib
     Diwan, A.A.; Peters, J.; Arenz, O. (2026). A Geometric Perspective on Reward Function Updates in Inverse Reinforcement Learning, Advances in Neural Information Processing Systems (NeurIPS).
2. •       Bib
     Diwan, A.A.; Tateo, D.; Mower, C.; Bou-Ammar, H.; Peters, J.; Arenz, O. (2026). Trust Region Inverse Reinforcement Learning: Explicit Dual Ascent using Local Policy Updates, International Conference on Machine Learning (ICML).
   •   Bib
     Diwan, A.A.; Tateo, D.; Mower, C.; Bou-Ammar, H.; Peters, J.; Arenz, O. (2026). Trust Region Inverse Reinforcement Learning: Explicit Dual Ascent using Local Policy Updates, European Workshop on Reinforcement Learning (EWRL).
3. •       Bib
     Diwan, A.A.; Urain, J.; Kober, J.; Peters, J. (2025). Noise-conditioned Energy-based Annealed Rewards (NEAR): A Generative Framework for Imitation Learning from Observation, International Conference on Learning Representations (ICLR).