Risk-Averse Model Uncertainty for Distributionally Robust Safe Reinforcement Learning

Authors: James Queeney, Mouhacine Benosman

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In experiments on continuous control tasks with safety constraints, we demonstrate that our framework produces robust performance and safety at deployment time across a range of perturbed test environments.
Researcher Affiliation Collaboration James Queeney Division of Systems Engineering Boston University jqueeney@bu.edu Mouhacine Benosman Mitsubishi Electric Research Laboratories benosman@merl.com
Pseudocode Yes Algorithm 1 Risk-Averse Model Uncertainty for Safe RL
Open Source Code Yes Code is publicly available at https://github.com/jqueeney/robust-safe-rl.
Open Datasets Yes we conduct experiments on 5 continuous control tasks with safety constraints from the Real-World RL Suite [18, 19]
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) for training, validation, and testing.
Hardware Specification Yes All experiments were run on a Linux cluster with 2.9 GHz Intel Gold processors and NVIDIA A40 and A100 GPUs.
Software Dependencies No The paper mentions software components like 'ELU activations' and 'multivariate Gaussian policy' but does not specify their version numbers or the versions of underlying deep learning frameworks or libraries.
Experiment Setup Yes Table 5: Network architectures and algorithm hyperparameters used in experiments