C3-UniMM: Causal Cycle-Consistent Unified Multimodal Modeling via Super Alignment and Shared Decoding Space
This paper introduces a Structured Latent Causal Graph (SLCG) as a shared cross-modal semantic space and proposes a Unified Decoding Space to enforce structural preservation and semantic invertibility during the cross-modal generation process.
Yu-Jie Shen, Lianlei Shan
· 0 citations