Essay.06-ZhancunMu-2100017790

Download PDF

Zhancun Mu

13 Nov 2023 (modified: 26 Jan 2024)PKU 2023 Fall CoRe SubmissionEveryoneRevisionsBibTeXCC BY 4.0
Keywords: GCRL, Goal representation
Abstract: Goal-conditioned reinforcement learning (GCRL) aims to design agents capable of solving multiple tasks. However, choosing, representing, and learning goals remains an open area of research. This essay first reviews goal representation and learning methods in GCRL. It then discusses challenges and future directions.
Submission Number: 138