发表机构
Linköping University; Uppsala University（林雪平大学; 乌普萨拉大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出交互粒子引导（IPG），通过传输替代重新加权，从奖励倾斜的生成先验中采样，克服SMC的权重退化与粒子坍缩，在图像修复和蛋白质结构推断任务上验证了有效性。
AI 中文摘要
推理时引导将预训练扩散模型和基于流的模型适配到新任务，例如从条件分布生成样本或生成具有期望属性的样本，而无需重新训练。这可以形式化为从奖励倾斜的生成先验中采样。由于从该分布精确采样不可行，基于引导的方法依赖近似产生有偏样本，而序贯蒙特卡洛（SMC）方法通过重要性权重纠正这种偏差。然而，尽管在大粒子极限下是精确的，SMC在实际中遭受权重退化与粒子坍缩问题。我们提出交互粒子引导（IPG），用传输替代重新加权。粒子通过一个额外漂移相互作用，该漂移源自Feynman–Kac偏微分方程以抵消重新加权项，并保持无权重。在再生核希尔伯特空间中选择漂移可得到闭式解，计算成本低，相比SMC开销可忽略。我们在具有已知后验的高斯混合模型，以及高维图像修复和蛋白质结构推断任务上展示了该方法。
英文摘要
Inference-time steering adapts pretrained diffusion and flow-based models to new tasks, e.g., to generate samples from a conditional distribution or samples with desired properties, without retraining. This can be formalized as sampling from a reward-tilted generative prior. As exact sampling from this distribution is intractable, guidance-based methods rely on approximations producing biased samples, and sequential Monte Carlo (SMC) methods correct for this bias using importance weights. However, while exact in the large particle limit, SMC suffers from weight degeneracy and particle collapse in practice. We propose interacting particle guidance (IPG), which replaces reweighting with transport. The particles interact through an additional drift, derived from the Feynman--Kac PDE to cancel the reweighting term, and remain unweighted. Choosing the drift in a reproducing kernel Hilbert space yields a closed-form solution that is cheap to compute, with negligible overhead compared to SMC. We demonstrate the method on Gaussian mixtures with known posteriors, and on high-dimensional image inpainting and protein structure inference tasks.
CommentsCode will be made available after publication
