Variance-Tilted Diffusion Models for Diverse Sampling
Iskander Azangulov, Leo Zhang, Kianoosh Ashouritaklimi
stat.ML, cs.LG
Submitted: 2026-06-20
Comments: Accepted at SPIGM @ ICML workshop 2026
License: http://creativecommons.org/licenses/by/4.0/
The gist: Diffusion models are typically sampled independently, even when the downstream objective is to obtain a diverse set of candidates.
Terminology
Abstract
Diffusion models are typically sampled independently, even when the downstream objective is to obtain a diverse set of candidates. We introduce a variance-weighted batch distribution that favours collections of samples with large empirical spread after a prescribed linear feature map. The target is specified explicitly, and the sampler is derived as the corresponding Doob h-transform of independent diffusion dynamics. The resulting correction has a compact form: an interaction term that repels posterior denoised means, together with a curvature term that moves particles to the region of higher feature variance. This yields an interacting-particle sampler with a transparent probabilistic target rather than a heuristic repulsive drift.
Sources
- Classifier-Free Diffusion Guidance
- SigmaDock: Untwisting Molecular Docking With Fragment-Based SE(3) Diffusion
- Score-Based Generative Modeling through Stochastic Differential Equations
Related papers
- Behavior of prediction performance metrics with rare events
- Optimal Estimation of Generic Dynamics by Path-Dependent Neural Jump ODEs
- A Posterior-Dynamics Framework for Imaging Inverse Problems with Pretrained Diffusion Priors
- Scalable Posterior Uncertainty for Flexible Density-Based Clustering
- A Transport Geometry of Benign Overfitting
- Efficient quadratic entropy with distance sketches