Group-Aware Coordination Graph for Multi-Agent Reinforcement Learning

Authors: Wei Duan, Jie Lu, Junyu Xuan

IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our evaluations, conducted on Star Craft II micromanagement tasks, demonstrate GACG s superior performance. An ablation study further provides experimental evidence of the effectiveness of each component of our method.
Researcher Affiliation Academia Wei Duan , Jie Lu , Junyu Xuan Australian Artificial Intelligence Institute (AAII), University of Technology Sydney wei.duan@student.uts.edu.au, {jie.lu, junyu.xuan}@uts.edu.au
Pseudocode No The paper describes the method using textual explanations and a framework diagram (Figure 2), but no pseudocode or explicitly labeled algorithm block is provided.
Open Source Code Yes The code is available at: https://github.com/Wei9711/GACG
Open Datasets Yes The experiments in this study are conducted using the Star Craft II benchmark [Samvelyan et al., 2019a]
Dataset Splits No The paper does not specify exact training, validation, and test dataset splits, nor does it describe a cross-validation setup. It mentions '5 random seeds' for experimental robustness but not for data partitioning.
Hardware Specification No The paper mentions 'GPUs' in a general context regarding parallelization but does not provide specific hardware details such as GPU or CPU models, processor types, or memory specifications used for the experiments.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies or libraries used in the implementation or experiments.
Experiment Setup Yes The environments are configured with a difficulty level of 7. The experiments are systematically carried out with 5 random seeds to ensure robustness and reliability in the assessment of the proposed methods... The training involves minimizing a loss function, composed of a temporal-difference (TD) loss and the group distance loss, as follows: L(θ) = LT D(θ ) + λLg (θg)... We explore different values for k, specifically {1, 5, 10, 20}.