AI RESEARCH
Compositional Transduction with Latent Analogies for Offline Goal-Conditioned Reinforcement Learning
arXiv CS.LG
•
ArXi:2605.20609v1 Announce Type: new Compositional generalization is essential for reaching unseen goals under novel contextual variations in offline goal-conditioned reinforcement learning (GCRL), where a generalist goal-reaching agent must be learned from limited data. Most prior approaches pursue this via trajectory stitching over temporally contiguous segments, which limits composing behaviors across varying contexts.