Efficient Wasserstein Natural Gradients for Reinforcement Learning

v1v2v3v4 (latest)

Efficient Wasserstein Natural Gradients for Reinforcement Learning

12 October 2020

Theodore H. Moskovitz

ArXiv (abs)PDF HTML

Papers citing "Efficient Wasserstein Natural Gradients for Reinforcement Learning"

19 / 19 papers shown

Title
Constrained Sliced Wasserstein Embedding Navid Naderializadeh Darian Salehi Hengrong Du Soheil Kolouri 106 4 0 02 Jun 2025
Wasserstein Policy Optimization David Pfau Ian Davies Diana Borsa Joao G. M. Araujo Brendan D. Tracey H. V. Hasselt 180 2 0 01 May 2025
A Conservative Approach for Few-Shot Transfer in Off-Dynamics Reinforcement Learning Paul Daoudi Christophe Prieur Bogdan Robu M. Barlier Ludovic Dos Santos OffRL 126 1 0 24 Dec 2023
Confronting Reward Model Overoptimization with Constrained RLHF Ted Moskovitz Aaditya K. Singh DJ Strouse Tuomas Sandholm Ruslan Salakhutdinov Anca D. Dragan Alexander Shmakov 179 67 0 06 Oct 2023
A State Representation for Diminishing RewardsNeural Information Processing Systems (NeurIPS), 2023 Ted Moskovitz Samo Hromadka Ahmed Touati Diana Borsa M. Sahani 89 2 0 07 Sep 2023
Wasserstein Diversity-Enriched Regularizer for Hierarchical Reinforcement LearningInternational Conference on Neural Information Processing (ICONIP), 2023 Haorui Li Jiaqi Liang Linjing Li D. Zeng 58 0 0 02 Aug 2023
Provably Convergent Policy Optimization via Metric-aware Trust Region Methods Jun Song Niao He Lijun Ding Chaoyue Zhao 167 3 0 25 Jun 2023
Wasserstein Gradient Flows for Optimizing Gaussian Mixture PoliciesNeural Information Processing Systems (NeurIPS), 2023 Hanna Ziesche Leonel Rozo 140 9 0 17 May 2023
The Geometry of Neural Nets' Parameter Spaces Under ReparametrizationNeural Information Processing Systems (NeurIPS), 2023 Agustinus Kristiadi Felix Dangel Philipp Hennig 136 16 0 14 Feb 2023
ReLOAD: Reinforcement Learning with Optimistic Ascent-Descent for Last-Iterate Convergence in Constrained MDPsInternational Conference on Machine Learning (ICML), 2023 Theodore H. Moskovitz Brendan O'Donoghue Vivek Veeriah Sebastian Flennerhag Satinder Singh Tom Zahavy 169 23 0 02 Feb 2023
Transfer RL via the Undo Maps Formalism Abhi Gupta Theodore H. Moskovitz David Alvarez-Melis Aldo Pacchiano OffRL 101 0 0 26 Nov 2022
Geometry and convergence of natural policy gradient methodsInformation Geometry (IG), 2022 Johannes Muller Guido Montúfar 143 13 0 03 Nov 2022
Learning General World Models in a Handful of Reward-Free DeploymentsNeural Information Processing Systems (NeurIPS), 2022 Yingchen Xu Jack Parker-Holder Aldo Pacchiano Philip J. Ball Oleh Rybkin Stephen J. Roberts Tim Rocktaschel Edward Grefenstette OffRL 175 10 0 23 Oct 2022
Trust Region Policy Optimization with Optimal Transport Discrepancies: Duality and Algorithm for Continuous ActionsNeural Information Processing Systems (NeurIPS), 2022 Antonio Terpin Nicolas Lanzetti Batuhan Yardim Florian Dorfler Giorgia Ramponi 92 8 0 20 Oct 2022
Minimum Description Length ControlInternational Conference on Learning Representations (ICLR), 2022 Theodore H. Moskovitz Ta-Chu Kao M. Sahani M. Botvinick 140 1 0 17 Jul 2022
Towards an Understanding of Default Policies in Multitask Policy OptimizationInternational Conference on Artificial Intelligence and Statistics (AISTATS), 2021 Theodore H. Moskovitz Michael Arbel Jack Parker-Holder Aldo Pacchiano 182 10 0 04 Nov 2021
A First-Occupancy Representation for Reinforcement Learning Theodore H. Moskovitz S. Wilson M. Sahani 205 16 0 28 Sep 2021
MICo: Improved representations via sampling-based state similarity for Markov decision processesNeural Information Processing Systems (NeurIPS), 2021 Pablo Samuel Castro Tyler Kastner Prakash Panangaden Mark Rowland 213 42 0 03 Jun 2021
Tactical Optimism and Pessimism for Deep Reinforcement LearningNeural Information Processing Systems (NeurIPS), 2021 Theodore H. Moskovitz Jack Parker-Holder Aldo Pacchiano Michael Arbel Sai Li 203 65 0 07 Feb 2021