ICML 2023 accepted papers on Image generation
76 of the 1,865 papers accepted at ICML 2023 (4.1%) are on Image generation, compared with 1.6% at ICML 2022. They fall into 7 topics.
See ICML 2023 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Fast diffusion sampling · 34 papers
- Consistency Models (23 citations)
- On Provable Copyright Protection for Generative Models (21 citations)
- Cones: Concept Neurons in Diffusion Models for Customized Generation (19 citations)
- Fast Sampling of Diffusion Models via Operator Learning (17 citations)
- Towards Coherent Image Inpainting Using Denoising Diffusion Implicit Models (17 citations)
- Reduce, Reuse, Recycle: Compositional Generation with Energy-Based Diffusion Models and MCMC (15 citations)
- Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models (13 citations)
- Are Diffusion Models Vulnerable to Membership Inference Attacks? (11 citations)
- PFGM++: Unlocking the Potential of Physics-Inspired Generative Models (11 citations)
- Input Perturbation Reduces Exposure Bias in Diffusion Models (10 citations)
- Dirichlet Diffusion Score Model for Biological Sequence Generation (8 citations)
- Diffusion Based Representation Learning (6 citations)
and 22 more in the interactive map.
Autoregressive and latent generation · 21 papers
- AudioLDM: Text-to-Audio Generation with Latent Diffusion Models (79 citations)
- SE(3) diffusion model with application to protein backbone generation (71 citations)
- TabDDPM: Modelling Tabular Data with Diffusion Models (66 citations)
- Composer: Creative and Controllable Image Synthesis with Composable Conditions (51 citations)
- Geometric Latent Diffusion Models for 3D Molecule Generation (31 citations)
- Text Generation with Diffusion Language Models: A Pre-training Approach with Continuous Paragraph Denoise (15 citations)
- Scalable Adaptive Computation for Iterative Generation (14 citations)
- MolDiff: Addressing the Atom-Bond Inconsistency Problem in 3D Molecule Diffusion Generation (11 citations)
- Hierarchical Neural Coding for Controllable CAD Model Generation (9 citations)
- AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners (7 citations)
- PLay: Parametrically Conditioned Layout Generation using Latent Diffusion (7 citations)
- Coarse-to-Fine: a Hierarchical Diffusion Model for Molecule Generation in 3D (5 citations)
and 9 more in the interactive map.
Image restoration · 9 papers
- Image Restoration with Mean-Reverting Stochastic Differential Equations (67 citations)
- I$^2$SB: Image-to-Image Schrödinger Bridge (15 citations)
- SinDDM: A Single Image Denoising Diffusion Model (9 citations)
- GibbsDDRM: A Partially Collapsed Gibbs Sampler for Solving Blind Inverse Problems with Denoising Diffusion Restoration (7 citations)
- Improving Statistical Fidelity for Neural Image Compression with Implicit Local Likelihood Models (7 citations)
- Restoration based Generative Models
- Fourmer: An Efficient Global Modeling Paradigm for Image Restoration
- IRNeXt: Rethinking Convolutional Network Design for Image Restoration
- Random Shuffle Transformer for Image Restoration
Text-to-image diffusion · 6 papers
- Muse: Text-To-Image Generation via Masked Generative Transformers (117 citations)
- MultiDiffusion: Fusing Diffusion Paths for Controlled Image Generation (63 citations)
- StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis (61 citations)
- Make-An-Audio: Text-To-Audio Generation with Prompt-Enhanced Diffusion Models (47 citations)
- One Transformer Fits All Distributions in Multi-Modal Diffusion at Scale (28 citations)
- Tractable Control for Autoregressive Language Generation (2 citations)
Super-resolution · 4 papers
- simple diffusion: End-to-end diffusion for high resolution images (27 citations)
- DeSRA: Detect and Delete the Artifacts of GAN-based Real-World Super-Resolution Models (11 citations)
- Crafting Training Degradation Distribution for the Accuracy-Generalization Trade-off in Real-World Super-Resolution (2 citations)
- Learning Controllable Degradation for Real-World Super-Resolution via Constrained Flows
GANs · 1 paper
Editing and style transfer · 1 paper
- Raising the Cost of Malicious AI-Powered Image Editing (15 citations)