MotionRL: Align Text-to-Motion Generation to Human Preferences with Multi-Reward Reinforcement Learning

We introduce MotionRL, the first approach to utilize Multi-Reward Reinforcement Learning (RL) for optimizing text-to-motion generation tasks and aligning them with human preferences. Previous works focused on improving numerical performance metrics on the given datasets, often neglecting the variabi...

Ausführliche Beschreibung

Bibliographische Detailangaben

Veröffentlicht in:	arXiv.org 2024-10
Hauptverfasser:	Liu, Xiaoyang, Mao, Yunyao, Zhou, Wengang, Li, Houqiang
Format:	Artikel
Sprache:	eng
Schlagworte:	Algorithms Human motion Human performance Multiple objective analysis Pareto optimization Performance measurement
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!