Variance-Tilted Diffusion Models for Diverse Sampling
This paper introduces a variance-weighted batch distribution to improve the diversity of samples in diffusion models.
This paper introduces a novel method to improve the diversity of samples in diffusion models by using a variance-weighted batch distribution.
Before reading this…
Applications
- →generative models
- →machine learning
To understand this paper, make sure you know these concepts first:
- diffusion modelsfind papers →
- Doob h-transformfind papers →
Abstract
More Like ThisDiffusion 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.