Abductive World Modeling via Causal Representation Learning
Abductive World Modeling (AWM), a framework that learns structured causal representations by abductively inferring latent causes from predicted futures, is proposed, which is the first framework to introduce abductive state inference into latent-space world modeling for learning structured representations of world dyna...
Zi-Qi Liu, Song-Han Yang, Lin-Fan Zhou et al.
· 0 citations