arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2306.04445cs.LGcs.CV
收藏

多模态潜变量扩散模型

Multi-modal Latent Diffusion

Renault Software Factory(雷诺软件工厂);EURECOM(欧洲通信学院)

Mustapha Bounoua, Giulio Franzese, Pietro Michiardi

AI总结:

针对多模态VAE的生成质量与跨模态连贯性权衡,本文提出用独立单模态确定性自编码器拼接潜空间并结合掩码扩散与多次条件分数训练,显著提升两项指标。

摘要:

多模态数据集在现代应用中无处不在,而多模态变分自编码器是一类广受欢迎的模型,旨在学习不同模态的联合表示。然而,现有方法存在连贯性—质量权衡问题:生成质量较好的模型缺乏跨模态的生成连贯性,反之亦然。我们讨论了现有方法性能不佳背后的局限性,以说明采用不同方法的必要性。我们提出了一种新颖方法,使用一组独立训练的单模态确定性自编码器。各个潜变量被拼接成一个公共潜空间,并输入给掩码扩散模型以实现生成建模。我们还引入了一种新的多次训练方法,用于学习多模态扩散的条件分数网络。大量实验表明,我们的方法在生成质量和连贯性两方面均显著优于竞争方法。

英文摘要:

Multi-modal data-sets are ubiquitous in modern applications, and multi-modal Variational Autoencoders are a popular family of models that aim to learn a joint representation of the different modalities. However, existing approaches suffer from a coherence-quality tradeoff, where models with good generation quality lack generative coherence across modalities, and vice versa. We discuss the limitations underlying the unsatisfactory performance of existing methods, to motivate the need for a different approach. We propose a novel method that uses a set of independently trained, uni-modal, deterministic autoencoders. Individual latent variables are concatenated into a common latent space, which is fed to a masked diffusion model to enable generative modeling. We also introduce a new multi-time training method to learn the conditional score network for multi-modal diffusion. Our methodology substantially outperforms competitors in both generation quality and coherence, as shown through an extensive experimental campaign.

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。

↑