方法对比

	多模态扩散模型	多模态视觉变换器
领域	深度学习	深度学习
方法族	Machine learning	Machine learning
起源年份	2020–2022	2021
提出者	Ho, J. et al. (DDPM); Rombach, R. et al. (LDM/Stable Diffusion)	Dosovitskiy et al. (ViT); Radford et al. (CLIP multimodal ViT)
类型	Generative model (denoising diffusion)	Multimodal transformer model
开创性文献	Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022). High-Resolution Image Synthesis with Latent Diffusion Models. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 10684–10695.	Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., & Houlsby, N. (2021). An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. In International Conference on Learning Representations (ICLR).
别名	multimodal DDPM, cross-modal diffusion, conditional multimodal diffusion, multi-modal denoising diffusion	Multimodal ViT, vision-language transformer, cross-modal vision transformer, multi-modal ViT
相关	6	5
摘要	A multimodal diffusion model extends denoising diffusion probabilistic models to generate or understand content by conditioning on signals from multiple modalities — such as text, image, audio, or video — simultaneously. It learns to reverse a noise process guided by cross-modal context, enabling high-fidelity synthesis and translation across modalities.	Multimodal Vision Transformer (Multimodal ViT) extends the Vision Transformer architecture to jointly process and align representations from multiple modalities — typically images and text — using self-attention and cross-attention mechanisms. By learning shared or aligned embedding spaces across modalities, it enables tasks such as visual question answering, image-text retrieval, visual grounding, and image captioning.