Learning Diverse Risk Preferences in Population-Based Self-Play

Authors: Yuhua Jiang, Qihan Liu, Xiaoteng Ma, Chenghao Li, Yiqin Yang, Jun Yang, Bin Liang, Qianchuan Zhao

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

Reproducibility Variable Result LLM Response
Research Type Experimental Our empirical results demonstrate that our method achieves comparable or superior performance in competitive games and, importantly, leads to the emergence of diverse behavioral modes.
Researcher Affiliation Academia Department of Automation, Tsinghua University {jiangyh22, lqh20, ma-xt17, lch18, yangyiqi19}@mails.tsinghua.edu.cn {yangjun603, bliang, zhaoqc}@tsinghua.edu.cn
Pseudocode Yes The algorithm is shown in Algorithm 1.
Open Source Code Yes Code is available at https://github.com/Jackory/RPBT.
Open Datasets Yes We consider two competitive multi-agent benchmarks: Slimevolley (Ha 2020) and Sumoants (Al-Shedivat et al. 2018).
Dataset Splits No The paper mentions training and testing but does not explicitly specify validation dataset splits (e.g., percentages or counts) or a clear methodology for a validation set.
Hardware Specification Yes All the experiments are conducted with one 64-core CPU and one Ge Force RTX 3090 GPU.
Software Dependencies No The paper mentions the use of PPO and other frameworks but does not provide specific version numbers for software dependencies or libraries.
Experiment Setup Yes We trained RPBT with population size 5 and set initial risk levels to {0.1, 0.4, 0.5, 0.6, 0.9} for all the experiments. For each method, we trained 3 runs using different random seeds and selected the one with the highest ELO score for evaluation.