Diverse Sampling in Diffusion Models with Marginal Preserving Particle Guidance

TL;DR

EDDY is a guidance method for diffusion models that enhances sample diversity by leveraging divergence-free dynamics, preserving marginal distributions without extra training.

Contribution

It introduces a novel divergence-free guidance mechanism based on Fokker-Planck symmetries, improving diversity in diffusion models without additional training.

Findings

EDDY increases diversity in synthetic and real-world image generation.

It maintains distributional fidelity comparable to baseline methods.

Practical approximations make EDDY computationally feasible.

Abstract

We present EDDY (Exact-marginal Diversification via Divergence-free dYnamics), a guidance mechanism for diffusion and flow matching models that promotes diversity among samples generated while maintaining quality. EDDY exploits symmetries of the Fokker-Planck equation, using drift perturbations that change particle trajectories while preserving the evolving marginal distribution. We instantiate this principle through kernel-based anti-symmetric pairwise matrix fields, constructed from the repulsive directions. The resulting divergence-free dynamics promote diversity at the joint particle level while preserving each particle's marginal distribution without any additional training. As computing the guidance can be computationally expensive in cases such as text-to-image generation with perceptual embeddings, we propose practical approximations as an effective and efficient solution. Experiments on synthetic distributions and text-to-image generation show that EDDY improves diversity while maintaining strong distributional fidelity compared to common baselines.