Variational Offline Multi-agent Skill Discovery

arXiv:2405.16386v1 Announce Type: new
Abstract: Skills are effective temporal abstractions established for sequential decision making tasks, which enable efficient hierarchical learning for long-horizon tasks and facilitate multi-task learning through their transferability. Despite extensive research, research gaps remain in multi-agent scenarios, particularly for automatically extracting subgroup coordination patterns in a multi-agent task. In this case, we propose two novel auto-encoder schemes: VO-MASD-3D and VO-MASD-Hier, to simultaneously capture subgroup- and temporal-level abstractions and form multi-agent skills, which firstly solves the aforementioned challenge. …

abstract abstractions agent arxiv case cs.ai cs.lg decision decision making discovery hierarchical horizon making multi-agent multi-task learning offline patterns research skill skills tasks temporal through type