Fusing shallower encoder layers into a Representation Autoencoder (RAE) sharpens pixel reconstruction but degrades diffusion generation. Prior studies and the research community commonly attribute this trade-off to the latent content itself, assuming detail-rich features are harder to model. We show that this gap stems...
Hong-Yang Du, Yun-Fei Xie, Jun-Jie Ye et al.· 1 citation
This work introduces FuseReg, which replaces heuristic feature selection with training over random subsets of encoder layers, and theoretically analyze the underlying mechanism: subset sampling preserves the full-layer latent mean in expectation while explicitly penalizing sensitivity to cross-layer disagreement.
Hong-Yang Du, Yun-Fei Xie, Jun-Jie Ye et al.· 1 citation
4DGS-JEPA is proposed, a Gaussian-native joint-embedding predictive architecture for causal multi-horizon prediction over dynamic Gaussian scenes and introduces a hybrid correspondence mechanism that combines persistent canonical identity with residual optimal-transport matching under reordering and topology change.
Joint-embedding predictive architectures learn abstract states by predicting target embeddings from context embeddings from context embeddings, but their transition models are typically opaque neural maps, so SJEPA is introduced, a reconstruction-free JEPA framework that learns predictive representations whose induced...
The introduction of Markov-Chain JEPA (MCJEPA), which replaces the latent predictor by a learned transition matrix, and Markov-Chain JEPA (MCJEPA), which replaces the latent predictor by a learned transition matrix, give temporal JEPA a principled state-space interpretation.
Yong-Chao Huang· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.