Flow Matching for Generative Modeling

Lipman, Yaron; Chen, Ricky T. Q.; Ben-Hamu, Heli; Nickel, Maximilian; Le, Maithao

doi:10.48550/arxiv.2210.02747

Cited by 14 publications

(23 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The ratio βt σ 2 t characterizes how fast the OT-ODE approaches X 0 , in a similar vein to the noise scheduler in SGM . With this interpretation in mind, we introduce our final result, which complements recent advances in flow-matching (Lipman et al, 2022) except for image-to-image problem setups. 2022)), which simply constructs a conditional score function with the newly available information (in this case, the degraded images) as an additional input.…”

Section: Connection To Flow-based Optimal Transport (Ot)mentioning

confidence: 95%

Quality assurance of image-guided radiation therapy

Liu

Nie

Cui

2023

Principles and Practice of Image-Guided Abdominal Radiation Therapy

View full text Add to dashboard Cite

We propose Image-to-Image Schrödinger Bridge (I 2 SB), a new class of conditional diffusion models that directly learn the nonlinear diffusion processes between two given distributions. These diffusion bridges are particularly useful for image restoration, as the degraded images are structurally informative priors for reconstructing the clean images. I 2 SB belongs to a tractable class of Schrödinger bridge, the nonlinear extension to score-based models, whose marginal distributions can be computed analytically given boundary pairs. This results in a simulation-free framework for nonlinear diffusions, where the I 2 SB training becomes scalable by adopting practical techniques used in standard diffusion models. We validate I 2 SB in solving various image restoration tasks, including inpainting, super-resolution, deblurring, and JPEG restoration on ImageNet 256×256 and show that I 2 SB surpasses standard conditional diffusion models with more interpretable generative processes. Moreover, I 2 SB matches the performance of inverse methods that additionally require the knowledge of the corruption operators. Our work opens up new algorithmic opportunities for developing efficient nonlinear diffusion models on a large scale. Project page: https://i2sb.github.io/.

show abstract

Section: Connection To Flow-based Optimal Transport (Ot)mentioning

confidence: 95%

Quality assurance of image-guided radiation therapy

Liu

Nie

Cui

2023

Principles and Practice of Image-Guided Abdominal Radiation Therapy

View full text Add to dashboard Cite

show abstract

“…Then generating graphics becomes very simple, as long as a point is sampled from a known simple distribution, and then through the inverse function of f , the picture can be generated. The Continuous Normalizing Flows (CNFs) [18] propose the concept of Flow Matching (FM), a simulation-free method for training CNFs based on regression vector fields with fixed conditional probability paths. Compared with Variational Autoencoder (VAE), the Normalizing Flow method [5] has achieved better performance in various image tasks.…”

Section: Related Workmentioning

confidence: 99%

Interpretable ODE-style Generative Diffusion Model via Force Field Construction

Jin¹,

Zhu²,

Peng³

2023

Preprint

View full text Add to dashboard Cite

For a considerable time, researchers have focused on developing a method that establishes a deep connection between the generative diffusion model and mathematical physics. Despite previous efforts, progress has been limited to the pursuit of a single specialized method. In order to advance the interpretability of diffusion models and explore new research directions, it is essential to establish a unified ODE-style generative diffusion model. Such a model should draw inspiration from physical models and possess a clear geometric meaning. This paper aims to identify various physical models that are suitable for constructing ODE-style generative diffusion models accurately from a mathematical perspective. We then summarize these models into a unified method. Additionally, we perform a case study where we use the theoretical model identified by our method to develop a range of new diffusion model methods, and conduct experiments. Our experiments on CIFAR-10 demonstrate the effectiveness of our approach. We have constructed a computational framework that attains highly proficient results with regards to image generation speed, alongside an additional model that demonstrates exceptional performance in both Inception score and FID score. These results underscore the significance of our method in advancing the field of diffusion models.

show abstract

“…where α(t) is a nonlinear function of t with a(0) = 1 and a(1) ≈ 0, and p(z) = N (0, I). Some recent work (Lipman et al, 2022; instead use the linear interpolation…”

Section: Background On Diffusion Modelsmentioning

confidence: 99%

A Study on the Tabloidization of News Posted on the Portal Site NAVER

Lee¹,

Kim²

2023

mgc

View full text Add to dashboard Cite

We present the Groupwise Diffusion Model (GDM), which divides data into multiple groups and diffuses one group at one time interval in the forward diffusion process. GDM generates data sequentially from one group at one time interval, leading to several interesting properties. First, as an extension of diffusion models, GDM generalizes certain forms of autoregressive models and cascaded diffusion models. As a unified framework, GDM allows us to investigate design choices that have been overlooked in previous works, such as data-grouping strategy and order of generation. Furthermore, since one group of the initial noise affects only a certain group of the generated data, latent space now possesses group-wise interpretable meaning. We can further extend GDM to the frequency domain where the forward process sequentially diffuses each group of frequency components. Dividing the frequency bands of the data as groups allows the latent variables to become a hierarchical representation where individual groups encode data at different levels of abstraction. We demonstrate several applications of such representation including disentanglement of semantic attributes, image editing, and generating variations.

show abstract

Flow Matching for Generative Modeling

Cited by 14 publications

References 13 publications

Quality assurance of image-guided radiation therapy

Quality assurance of image-guided radiation therapy

Interpretable ODE-style Generative Diffusion Model via Force Field Construction

A Study on the Tabloidization of News Posted on the Portal Site NAVER

Contact Info

Product

Resources

About