Diffuse Everything: Multimodal Diffusion Models on Arbitrary State Spaces
This addresses the challenge of generating coupled multimodal data without relying on external encoders/decoders, which is beneficial for applications with limited data, though it appears incremental as it builds on existing diffusion model foundations.
The paper tackles the problem of joint multimodal data generation with diffusion models, which typically require preprocessing to unify data formats, by proposing a framework that works directly on arbitrary state spaces with decoupled noise schedules. The result is a single model capable of unconditional and modality-conditioned generation, achieving competitive performance in text-image generation and mixed-type tabular data synthesis.
Diffusion models have demonstrated remarkable performance in generating unimodal data across various tasks, including image, video, and text generation. On the contrary, the joint generation of multimodal data through diffusion models is still in the early stages of exploration. Existing approaches heavily rely on external preprocessing protocols, such as tokenizers and variational autoencoders, to harmonize varied data representations into a unified, unimodal format. This process heavily demands the high accuracy of encoders and decoders, which can be problematic for applications with limited data. To lift this restriction, we propose a novel framework for building multimodal diffusion models on arbitrary state spaces, enabling native generation of coupled data across different modalities. By introducing an innovative decoupled noise schedule for each modality, we enable both unconditional and modality-conditioned generation within a single model simultaneously. We empirically validate our approach for text-image generation and mixed-type tabular data synthesis, demonstrating that it achieves competitive performance.