A survey on diffusion models for inverse problems

G Daras, H Chung, CH Lai, Y Mitsufuji, JC Ye… - arxiv preprint arxiv …, 2024 - arxiv.org
Diffusion models have become increasingly popular for generative modeling due to their
ability to generate high-quality samples. This has unlocked exciting new possibilities for …

Tfg: Unified training-free guidance for diffusion models

H Ye, H Lin, J Han, M Xu, S Liu, Y Liang, J Ma… - arxiv preprint arxiv …, 2024 - arxiv.org
Given an unconditional diffusion model and a predictor for a target property of interest (eg, a
classifier), the goal of training-free guidance is to generate samples with desirable target …

Sg-i2v: Self-guided trajectory control in image-to-video generation

K Namekata, S Bahmani, Z Wu, Y Kant… - arxiv preprint arxiv …, 2024 - arxiv.org
Methods for image-to-video generation have achieved impressive, photo-realistic quality.
However, adjusting specific elements in generated videos, such as object motion or camera …

Infusion: Preventing customized text-to-image diffusion from overfitting

W Zeng, Y Yan, Q Zhu, Z Chen, P Chu, W Zhao… - Proceedings of the …, 2024 - dl.acm.org
Text-to-image (T2I) customization aims to create images that embody specific visual
concepts delineated in textual descriptions. However, existing works still face a main …

Guidance with spherical gaussian constraint for conditional diffusion

L Yang, S Ding, Y Cai, J Yu, J Wang, Y Shi - arxiv preprint arxiv …, 2024 - arxiv.org
Recent advances in diffusion models attempt to handle conditional generative tasks by
utilizing a differentiable loss function for guidance without the need for additional training …

Diffusion models as constrained samplers for optimization with unknown constraints

L Kong, Y Du, W Mu, K Neklyudov, V De Bortoli… - arxiv preprint arxiv …, 2024 - arxiv.org
Addressing real-world optimization problems becomes particularly challenging when
analytic objective functions or constraints are unavailable. While numerous studies have …

No training, no problem: Rethinking classifier-free guidance for diffusion models

S Sadat, M Kansy, O Hilliges, RM Weber - arxiv preprint arxiv:2407.02687, 2024 - arxiv.org
Classifier-free guidance (CFG) has become the standard method for enhancing the quality
of conditional diffusion models. However, employing CFG requires either training an …

G2D2: Gradient-guided Discrete Diffusion for image inverse problem solving

N Murata, CH Lai, Y Takida, T Uesaka… - arxiv preprint arxiv …, 2024 - arxiv.org
Recent literature has effectively utilized diffusion models trained on continuous variables as
priors for solving inverse problems. Notably, discrete diffusion models with discrete latent …

Infusion: Preventing customized text-to-image diffusion from overfitting

Z Weili, Y Yan, Q Zhu, Z Chen, P Chu, W Zhao… - ACM Multimedia …, 2024 - openreview.net
Text-to-image (T2I) customization aims to create images that embody specific visual
concepts delineated in textual descriptions. However, existing works still face a main …

Gradient-Guided Conditional Diffusion Models for Private Image Reconstruction: Analyzing Adversarial Impacts of Differential Privacy and Denoising

T Huang, J Meng, H Chen, G Zheng, X Yang… - arxiv preprint arxiv …, 2024 - arxiv.org
We investigate the construction of gradient-guided conditional diffusion models for
reconstructing private images, focusing on the adversarial interplay between differential …