The Trade-off Between Covariate Dependence and Latent Structure in Representation Learning
A unified, supervised framework that couples latent dimension-covariate dependence with constraints on the latent structure is introduced, where enforcing latent independence or exclusive one-to-one latent-covariate dependence comes at a provable cost in latent-covariate alignment.