Asynchronous Actor-Critic for Multi-Agent Reinforcement Learning

Authors: Yuchen Xiao, Weihao Tan, Christopher Amato

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical results (in simulation and hardware) in a variety of realistic domains demonstrate the superiority of our approaches in large multi-agent problems and validate the effectiveness of our algorithms for learning high-quality and asynchronous solutions.
Researcher Affiliation Academia Yuchen Xiao Khoury College of Computer Sciences Northeastern University Boston, MA 02115 xiao.yuch@northeastern.edu Weihao Tan Khoury College of Computer Sciences Northeastern University Boston, MA 02115 w.tan@northeastern.edu Christopher Amato Khoury College of Computer Sciences Northeastern University Boston, MA 02115 c.amato@northeastern.edu
Pseudocode Yes The pseudocode and detailed trajectory squeezing process for each proposed method are presented in Appendix C.
Open Source Code Yes In supplementary materials, we include the code and a README.txt file to reproduce the main experimental results.
Open Datasets Yes We investigate the performance of our algorithms over a variety of multi-agent problems with macroactions (Fig. 1): Box Pushing [Xiao et al., 2019], Overcooked [Wu et al., 2021b], and a larger Warehouse Tool Delivery [Xiao et al., 2019] domain.
Dataset Splits No The paper refers to 'training trials' and 'testing episodes' for evaluation but does not specify explicit dataset splits (e.g., percentages or counts for training, validation, and test sets).
Hardware Specification Yes The details of used computational resources are mentioned in Appendix E.
Software Dependencies No The provided text does not explicitly list software dependencies with specific version numbers.
Experiment Setup Yes All the training details including hyperparameters are in Appendix E.