arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2310.04041cs.LGcs.AI
收藏

观测引导的扩散概率模型

Observation-Guided Diffusion Probabilistic Models

LG AI Research(LG AI研究院);Seoul National University(首尔大学)

Junoh Kang, Jinyoung Choi, Sungik Choi, Bohyung Han

AI总结:

本文提出观测引导的扩散概率模型(OGDM),通过引入基于噪声水平条件判别器的观测引导损失项,在保持快速采样的同时提升生成质量,且不增加推理计算成本。

摘要:

我们提出了一种新颖的基于扩散的图像生成方法,称为观测引导的扩散概率模型(OGDM),它有效地解决了质量控制与快速采样之间的权衡问题。我们的方法通过以有原则的方式将观测过程的引导与马尔可夫链相结合,重新建立了训练目标。这是通过引入一个额外的损失项来实现的,该损失项基于一个以噪声水平为条件的判别器从观测中推导出来,该判别器采用伯努利分布来指示其输入是否位于(含噪的)真实流形上。这一策略使我们能够在推理阶段优化更准确的负对数似然,尤其是在函数评估次数有限的情况下。所提出的训练方案即使仅被纳入微调过程中也具有优势,并且它与各种快速推理策略兼容,因为我们的方法在完全相同的推理过程中产生更好的去噪网络,而不会产生额外的计算成本。我们使用多种推理技术在强扩散模型基线上证明了我们训练算法的有效性。我们的实现可在 https://github.com/Junoh-Kang/OGDM_edm 获取。

英文摘要:

We propose a novel diffusion-based image generation method called the observation-guided diffusion probabilistic model (OGDM), which effectively addresses the tradeoff between quality control and fast sampling. Our approach reestablishes the training objective by integrating the guidance of the observation process with the Markov chain in a principled way. This is achieved by introducing an additional loss term derived from the observation based on a conditional discriminator on noise level, which employs a Bernoulli distribution indicating whether its input lies on the (noisy) real manifold or not. This strategy allows us to optimize the more accurate negative log-likelihood induced in the inference stage especially when the number of function evaluations is limited. The proposed training scheme is also advantageous even when incorporated only into the fine-tuning process, and it is compatible with various fast inference strategies since our method yields better denoising networks using the exactly the same inference procedure without incurring extra computational cost. We demonstrate the effectiveness of our training algorithm using diverse inference techniques on strong diffusion model baselines. Our implementation is available at https://github.com/Junoh-Kang/OGDM_edm.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑