SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2204.08957 · ICLR · 2022

COptiDICE: Offline Constrained Reinforcement Learning via Stationary Distribution Correction Estimation

Arthur Guez, Doina Precup, Nicolas Heess, Kaist, Daniel Mankowitz, Cosmin Paduraru, Kee-Eung Kim, Jongmin Lee

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 7 functions out of this paper's own repositories and ran 4 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
deepmind/constrained_optidice canonical 4 of 7
FunctionStatusWhere it lives
domain_cost_fn Ran deepmind/constrained_optidice/neural/cost.py
code served (permissive licence) · get_code("021d008f035670a2")
get_f_divergence_fn Ran deepmind/constrained_optidice/neural/learning.py
code served (permissive licence) · get_code("b0a316d25b4cad24")
policy_evaluation Ran deepmind/constrained_optidice/tabular/mdp_util.py
code served (permissive licence) · get_code("77a7c8ddd9479170")
policy_evaluation_mdp Ran deepmind/constrained_optidice/tabular/mdp_util.py
code served (permissive licence) · get_code("3bb97e9b8f8bef6d")
conditional_update Not yet run deepmind/constrained_optidice/neural/learning.py
code served (permissive licence) · get_code("68bce47fe9bdd627")
generate_random_cmdp Not yet run deepmind/constrained_optidice/tabular/mdp_util.py
code served (permissive licence) · get_code("06a80085f44b4edd")
periodic_update Not yet run deepmind/constrained_optidice/neural/learning.py
code served (permissive licence) · get_code("ba29d1b163a7d24d")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We consider the offline constrained reinforcement learning (RL) problem, in which the agent aims to compute a policy that maximizes expected return while satisfying given cost constraints, learning only from a pre-collected dataset. This problem setting is appealing in many real-world scenarios, where direct interaction with the environment is costly or risky, and where the resulting policy should comply with safety constraints. However, it is challenging to compute a policy that guarantees satisfying the cost constraints in the offline RL setting, since the offpolicy evaluation inherently has an estimation error. In this paper, we present an offline constrained RL algorithm that optimizes the policy in the space of the stationary distribution. Our algorithm, COptiDICE, directly estimates the stationary distribution corrections of the optimal policy with respect to returns, while constraining the cost upper bound, with the goal of yielding a cost-conservative policy for actual constraint satisfaction. Experimental results show that COptiDICE attains better policies in terms of constraint satisfaction and return-maximization, outperforming baseline algorithms.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2204.08957")
get_code_for_paper("2204.08957")
have("2204.08957")

Connect an agent — have() is free.