AI RESEARCH
Joint Learning of Hierarchical Neural Options and Abstract World Model
arXiv CS.AI
•
ArXi:2602.02799v2 Announce Type: replace-cross Building agents that can perform new skills by composing existing skills is a long-standing goal of AI agent research. Towards this end, we investigate how to efficiently acquire a sequence of skills, formalized as hierarchical neural options. However, existing model-free hierarchical reinforcement algorithms need a lot of data.