SOURCE-LINKED INTELLIGENCE

Beyond Random Couplings: Contrastive Noise Alignment in Generative Flows

arXiv · AI, language, vision and robotics · article · Sep 16, 2026 · UTC

Diffusion and flow-matching models are typically trained by corrupting data through independently sampled Gaussian noise. While simple and scalable, this forward process induces arbitrary data-noise couplings, forcing the network to learn high-curvature transports between unrelated endpoints. Existing optimal-transport methods reduce this burden by reassigning fixed noise samples to data, but the source noise distribution itself remains passive. To address this, we introduce Contrastive Noise Alignment (CNA), a training-time method that creates dynamic, contrastive couplings by optimizing the

Read original source ↗ Open in workspace

recordType

paper

region

Global