arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2204.00227cs.CVcs.LG
收藏

扩散模型的感知优先训练

Perception Prioritized Training of Diffusion Models

Seoul National University(首尔国立大学);LG AI Research(LG人工智能研究院)

Jooyoung Choi, Jungbeom Lee, Chaehun Shin, Sungwon Kim, Hyunwoo Kim, Sungroh Yoon

摘要:

扩散模型通过优化相应损失项的加权和来学习恢复被不同级别的噪声破坏的噪声数据,即,去噪得分匹配损失。在本文中,我们表明,恢复数据损坏一定的噪声水平提供了一个适当的借口任务的模型学习丰富的视觉概念。我们建议通过重新设计目标函数的加权方案,在训练过程中优先考虑这些噪声水平。我们表明,我们简单的重新设计的加权方案显着提高性能的扩散模型,无论数据集,架构和采样策略。

英文摘要:

Diffusion models learn to restore noisy data, which is corrupted with different levels of noise, by optimizing the weighted sum of the corresponding loss terms, i.e., denoising score matching loss. In this paper, we show that restoring data corrupted with certain noise levels offers a proper pretext task for the model to learn rich visual concepts. We propose to prioritize such noise levels over other levels during training, by redesigning the weighting scheme of the objective function. We show that our simple redesign of the weighting scheme significantly improves the performance of diffusion models regardless of the datasets, architectures, and sampling strategies.

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。

↑