Universality of Gaussian-Mixture Reverse Kernels in Conditional Diffusion

arXiv:2604.13470v1 Announce Type: new Abstract: We prove that conditional diffusion models whose reverse kernels are finite Gaussian mixtures with ReLU-network logits can approximate suitably regular target distributions arbitrarily well in context-averaged conditional KL divergence, up to an irreducible terminal mismatch that typically vanishes with increasing diffusion horizon. A path-space decomposition reduces the output error to this mismatch plus per-step reverse-kernel errors; assuming each reverse kernel factors through a finite-dimensional feature map, each step becomes a static conditional density approximation problem, solved by composing Norets' Gaussian-mixture theory with quantitative ReLU bounds. Under exact terminal matching the resulting neural reverse-kernel class is dense in conditional KL.

Related

• Identifiability of Potentially Degenerate Gaussian Mixture Models With Piecewise Affine Mixing
• Signal-Aware Conditional Diffusion Surrogates for Transonic Wing Pressure Prediction
• Learning Discrete Diffusion of Graphs via Free-Energy Gradient Flows
• Adjoint Matching through the Lens of the Stochastic Maximum Principle in Optimal Control
• BezierFlow: Learning Bezier Stochastic Interpolant Schedulers for Few-Step Generation

Source: arXiv cs.LG | 2026-04-16