← Back to all news
arXiv Machine Learning

Disentangling Total-Variance and Signal-to-Noise-Ratio Improves Diffusion Models

The paper introduces a total‑variance/signal‑to‑noise‑ratio (TV/SNR) framework that allows independent control of TV and SNR in diffusion models. By replacing exponentially exploding TV schedules with constant TV while keeping the SNR schedule, the authors achieve consistent improvements in generation efficiency. The approach generalizes the SNR schedule from optimal transport flow matching and works across various reverse diffusion solvers, reducing the search space to two interpretable parameters for robust sampling in scientific applications such as molecular structure generation.

Related stories

arXiv Machine Learning
Sep 17

Beyond Random Couplings: Contrastive Noise Alignment in Generative Flows

The paper introduces Contrastive Noise Alignment (CNA), a training-time method for generative flow models that dynamically aligns Gaussian noise with data samples using a cross-modal InfoNCE objective. By modeling noise as an interacting particle system and regularizing with angular entropy and radial norm penalties, CNA reduces arbitrary data-noise couplings and flow curvature. Empirical results show that CNA improves generation quality, lowering FID by over 50% for few-step pixel-space generation compared to standard rectified flow and outperforming optimal transport baselines by at least 24%.

By Lennart Wittke, Vinicius Azevedo
diffusionmultimodalsafety
More like this →
arXiv Computer Vision
Sep 15

CrossDistill: Balancing Quality and Diversity via Trajectory-Level Hybrid Few-Step Distillation

CrossDistill is a trajectory-level hybrid few-step distillation framework for diffusion models that balances quality and diversity by splitting the sampling trajectory at a crossover point. The high-noise interval uses a trajectory-preserving objective to maintain global mode coverage, while the low-noise interval applies a distribution-matching objective to sharpen local details, with the two stages coupled through the crossover state. This noise-level scheduling policy, demonstrated on text-to-video and image-to-video diffusion models, expands the few-step quality-diversity frontier by preserving seed-level variation while achieving competitive visual fidelity.

By Yuxi Liu, Haoyu Li, Yixiang Cai, Tengxu Sun, Zekun Zhang, Baole Ai, Ang Wang, Jiamang Wang, Lin Qu, Kun Yuan, Kai Zhang
diffusionefficiency
More like this →
arXiv Machine Learning
Sep 3

Generative Diffusion Surrogates with Analytical Variance Schedule

The paper introduces a method for creating generative diffusion surrogates that use an analytically derived variance schedule based on the time derivative of known variance from macroscopic theory or empirical scaling. By enforcing this variance path, the model can capture non‑Gaussian structure in transport systems without requiring intermediate‑time physical data. The approach is validated on ballistic‑to‑diffusive transport in turbulent plasmas, accurately reproducing test‑particle distributions, laboratory‑measured variance, and simulated kurtosis evolution.

By Patrick Reichherzer, Gianluca Gregori, David N. Hosking, Subir Sarkar
diffusion
More like this →
arXiv Machine Learning
Sep 2

Control Variate Score Matching for Diffusion Models

arXiv:2512.20003v2 Announce Type: replace Abstract: Sampling from unnormalized probability densities is a pervasive challenge across the computational and physical sciences. Diffusion models provide...

By Khaled Kahouli, Romuald Elie, Klaus-Robert M\"uller, Quentin Berthet, Oliver T. Unke, Arnaud Doucet
diffusion
More like this →