Explore 3D Dance Generation via Reward Model from Automatically-Ranked Demonstrations
Abstract
This paper presents an Exploratory 3D Dance generation framework, E3D2, designed to address the exploration capability deficiency in existing music-conditioned 3D dance generation models. Current models often generate monotonous and simplistic dance sequences that misalign with human preferences because they lack exploration capabilities. The E3D2 framework involves a reward model trained from automatically-ranked dance demonstrations, which then guides the reinforcement learning process. This approach encourages the agent to explore and generate high quality and diverse dance movement sequences. The soundness of the reward model is both theoretically and experimentally validated. Empirical experiments demonstrate the effectiveness of E3D2 on the AIST++ dataset.
Cite
Text
Wang et al. "Explore 3D Dance Generation via Reward Model from Automatically-Ranked Demonstrations." AAAI Conference on Artificial Intelligence, 2024. doi:10.1609/AAAI.V38I1.27783Markdown
[Wang et al. "Explore 3D Dance Generation via Reward Model from Automatically-Ranked Demonstrations." AAAI Conference on Artificial Intelligence, 2024.](https://mlanthology.org/aaai/2024/wang2024aaai-explore/) doi:10.1609/AAAI.V38I1.27783BibTeX
@inproceedings{wang2024aaai-explore,
title = {{Explore 3D Dance Generation via Reward Model from Automatically-Ranked Demonstrations}},
author = {Wang, Zilin and Zhuang, Haolin and Li, Lu and Zhang, Yinmin and Zhong, Junjie and Chen, Jun and Yang, Yu and Tang, Boshi and Wu, Zhiyong},
booktitle = {AAAI Conference on Artificial Intelligence},
year = {2024},
pages = {301-309},
doi = {10.1609/AAAI.V38I1.27783},
url = {https://mlanthology.org/aaai/2024/wang2024aaai-explore/}
}