#artificial intelligence
Jun 2026
ConTraIRL: Factorized Contrastive Abstractions for Transferable IRL
Factorized Contrastive Abstractions for Transferable IRL (ConTraIRL), a framework that enables compositional reward transfer by learning decoupled latent representations of these two factors, is proposed.
Yi-Kang Gui, Bikramjit Banerjee, Prashant Doshi
· arXiv.org · 0 citations