Diffusion Model

Diffusion models are generative models that create data by reversing a noise-diffusion process, aiming to generate high-quality samples from complex distributions. Current research focuses on improving efficiency through techniques like stochastic Runge-Kutta methods and dynamic model architectures (e.g., Dynamic Diffusion Transformer), as well as enhancing controllability and safety via methods such as classifier-free guidance and reinforcement learning from human feedback. These advancements are significantly impacting various fields, including medical imaging, robotics, and artistic creation, by enabling novel applications in image generation, inverse problem solving, and multi-modal data synthesis.

Papers

October 12, 2024

DuoDiff: Accelerating Diffusion Models with a Dual-Backbone Approach
Daniel Gallo Fernández, Răzvan-Andrei Matişan, Alejandro Monroy Muñoz, Ana-Maria Vasilcoiu, Janusz Partyka, Tin Hadži Veljković, Metod Jazbec
Neural Network Diffusion Model Denoising Network Multi Exit Backbone Architecture
Enhancing Single Image to 3D Generation using Gaussian Splatting and Hybrid Diffusion Priors
Hritam Basak, Hadi Tabatabaee, Shreekant Gayaka, Ming-Feng Li, Xin Yang, Cheng-Hao Kuo, Arnie Sen, Min Sun, Zhaozheng Yin
Diffusion Model Gaussian Splatting Single Image 3D Generation 3D Diffusion Model 3D Structure 3D Object Generation

October 11, 2024

TD-Paint: Faster Diffusion Inpainting Through Time Aware Pixel Conditioning
Tsiry Mayet, Pourya Shamsolmoali, Simon Bernard, Eric Granger, Romain Hérault, Clement Chatelain
Diffusion Model Consistent Conditioning State of the Art Diffusion Faster Diffusion
Linear Convergence of Diffusion Models Under the Manifold Hypothesis
Peter Potaptchik, Iskander Azangulov, George Deligiannidis
Diffusion Model Generative Model Convergence Guarantee Linear Convergence Manifold Hypothesis Backward Stochastic Differential Equation Step Diffusion Model
Gait Sequence Upsampling using Diffusion Models for Single LiDAR Sensors
Jeongho Ahn, Kazuto Nakashima, Koki Yoshino, Yumi Iwashita, Ryo Kurazume
Diffusion Model Point Cloud Gait Recognition Single LiDAR Human Point Cloud
Diffusion Models Need Visual Priors for Image Generation
Xiaoyu Yue, Zidong Wang, Zeyu Lu, Shuyang Sun, Meng Wei, Wanli Ouyang, Lei Bai, Luping Zhou
Diffusion Model Image Generation Diffusion Explainer Diffusion Sampling Visual Prior

October 10, 2024

Avoiding mode collapse in diffusion models fine-tuned with reinforcement learning
Roberto Barceló, Cristóbal Alcázar, Felipe Tobar
Reinforcement Learning LeArning Abstract Diffusion Model Positive Reinforcement Diffusion Policy Robust Fine Tuning Mode Collapse Reward Fine Tuning
Dynamics of Concept Learning and Compositional Generalization
Yongyi Yang, Core Francisco Park, Ekdeep Singh Lubana, Maya Okawa, Wei Hu, Hidenori Tanaka
Diffusion Model Strong Generalization Compositional Generalization Concept Learning
DICE: Discrete Inversion Enabling Controllable Editing for Multinomial Diffusion and Masked Generative Models
Xiaoxiao He, Ligong Han, Quan Dao, Song Wen, Minhao Bai, Di Liu, Han Zhang, Martin Renqiang Min, Felix Juefei-Xu, Chaowei Tan, Bo Liu, Kang Li, Hongdong Li, Junzhou Huang, Faez Ahmed, Akash Srivastava, Dimitris Metaxas
Diffusion Model Generative Model Discrete Diffusion Model Controllable Editing Masked Generative Improved Box Dice
Meissonic: Revitalizing Masked Generative Transformers for Efficient High-Resolution Text-to-Image Synthesis
Jinbin Bai, Tian Ye, Wei Chow, Enxin Song, Qing-Guo Chen, Xiangtai Li, Zhen Dong, Lei Zhu, Shuicheng Yan
Diffusion Model High Efficiency Image Generation Text to Image Synthesis Image Fidelity Autoregressive Image Generation Masked Generative
DART: Denoising Autoregressive Transformer for Scalable Text-to-Image Generation
Jiatao Gu, Yuyang Wang, Yizhe Zhang, Qihang Zhang, Dinghuai Zhang, Navdeep Jaitly, Josh Susskind, Shuangfei Zhai
Diffusion Model Image Generation Image Synthesis Image Modeling Autoregressive Transformer Text to Image Generation Task Large Scale Text to Image
Unstable Unlearning: The Hidden Risk of Concept Resurgence in Diffusion Models
Vinith M. Suriyakumar, Rohan Alur, Ayush Sekhari, Manish Raghavan, Ashia C. Wilson
Diffusion Model Text to Image Diffusion Model Concept Erasure Hidden Risk Adversarial Information
Generated Bias: Auditing Internal Bias Dynamics of Text-To-Image Generative Models
Abhishek Mandal, Susan Leavy, Suzanne Little
Diffusion Model Text to Image Text to Image Generative Model Source Bias Bias Auditing
$\textit{Jump Your Steps}$: Optimizing Sampling Schedule of Discrete Diffusion Models
Yong-Hyun Park, Chieh-Hsin Lai, Satoshi Hayakawa, Yuhta Takida, Yuki Mitsufuji
Diffusion Model Fast Sampling Sample STEP Discrete Diffusion Model Sampling Technique Parallel Sampling
Data Augmentation for Surgical Scene Segmentation with Anatomy-Aware Diffusion Models
Danush Kumar Venkatesh, Dominik Rivoir, Micha Pfeiffer, Fiona Kolbinger, Stefanie Speidel
Diffusion Model Synthetic Data Surgical Datasets Target Mask

October 9, 2024

InstructG2I: Synthesizing Images from Multimodal Attributed Graphs
Bowen Jin, Ziqi Pang, Bingjun Guo, Yu-Xiong Wang, Jiaxuan You, Jiawei Han
Diffusion Model Graph Encoder Graph Node MPT 7b Instruct Multimodal Graph Image to Graph
Diffusion Density Estimators
Akhil Premkumar
Diffusion Model Density Estimation Generative Process Probability Flow ODE Neural Density Tractable Density
Jointly Generating Multi-view Consistent PBR Textures using Collaborative Control
Shimon Vainer, Konstantin Kutsy, Dante De Nigris, Ciara Rowles, Slava Elizarov, Simon Donné
Diffusion Model Pixel Wise Image Diffusion Model View Consistency Multi View Image Generation
Diff-FMT: Diffusion Models for Fluorescence Molecular Tomography
Qianqian Xue, Peng Zhang, Xingyu Liu, Wenjian Wang, Guanglei Zhang
Diffusion Model Image Reconstruction Optical Tomography Fluorescence Lifetime Fluorescence Spectroscopy
Suppress Content Shift: Better Diffusion Features via Off-the-Shelf Generation Techniques
Benyuan Meng, Qianqian Xu, Zitai Wang, Zhiyong Yang, Xiaochun Cao, Qingming Huang
Diffusion Model Pre Trained Diffusion Model Diffusion Feature

Diffusion Model

Papers

DuoDiff: Accelerating Diffusion Models with a Dual-Backbone Approach

Enhancing Single Image to 3D Generation using Gaussian Splatting and Hybrid Diffusion Priors

TD-Paint: Faster Diffusion Inpainting Through Time Aware Pixel Conditioning

Linear Convergence of Diffusion Models Under the Manifold Hypothesis

Gait Sequence Upsampling using Diffusion Models for Single LiDAR Sensors

Diffusion Models Need Visual Priors for Image Generation

Avoiding mode collapse in diffusion models fine-tuned with reinforcement learning

Dynamics of Concept Learning and Compositional Generalization

DICE: Discrete Inversion Enabling Controllable Editing for Multinomial Diffusion and Masked Generative Models

Meissonic: Revitalizing Masked Generative Transformers for Efficient High-Resolution Text-to-Image Synthesis

DART: Denoising Autoregressive Transformer for Scalable Text-to-Image Generation

Unstable Unlearning: The Hidden Risk of Concept Resurgence in Diffusion Models

Generated Bias: Auditing Internal Bias Dynamics of Text-To-Image Generative Models

$\textit{Jump Your Steps}$: Optimizing Sampling Schedule of Discrete Diffusion Models

Data Augmentation for Surgical Scene Segmentation with Anatomy-Aware Diffusion Models

InstructG2I: Synthesizing Images from Multimodal Attributed Graphs

Diffusion Density Estimators

Jointly Generating Multi-view Consistent PBR Textures using Collaborative Control

Diff-FMT: Diffusion Models for Fluorescence Molecular Tomography

Suppress Content Shift: Better Diffusion Features via Off-the-Shelf Generation Techniques