AGAN is the first neural architecture search method for GANs that discovers architectures outperforming state-of-the-art on CIFAR-10 unsupervised image generation and competitive on supervised tasks.
hub Mixed citations
Improved Techniques for Training GANs
Mixed citation behavior. Most common role is background (60%).
21
Pith papers citing it
1,386
external citations · Pith
Background
60% of classified citations
abstract
We present a variety of new architectural features and training procedures that we apply to the generative adversarial networks (GANs) framework. We focus on two applications of GANs: semi-supervised learning, and the generation of images that humans find visually realistic. Unlike most work on generative models, our primary goal is not to train a model that assigns high likelihood to test data, nor do we require the model to be able to learn well without using any labels. Using our new techniques, we achieve state-of-the-art results in semi-supervised classification on MNIST, CIFAR-10 and SVHN. The generated images are of high quality as confirmed by a visual Turing test: our model generates MNIST samples that humans cannot distinguish from real data, and CIFAR-10 samples that yield a human error rate of 21.3%. We also present ImageNet samples with unprecedented resolution and show that our methods enable the model to learn recognizable features of ImageNet classes.
hub tools
citation-role summary
background 3
method 2
citation-polarity summary
representative citing papers
Matched benchmarking reveals FID misleads in few-step regimes under CFG, prompting CLIP-scaled and PickScore-scaled FID and IS variants for better semantic evaluation of one-step image generators.
A single DiT-based diffusion model unifies video-to-audio, text-to-audio, and joint video-text-to-audio generation, supported by a new 470k-pair dataset and three-stage progressive training that resolves task competition.
Diffusion models with architecture improvements and classifier guidance achieve superior FID scores to GANs on unconditional and conditional ImageNet image synthesis.
FP-MGMs with consistency loss and three-state reuse (CoFRe) reduce parameters by up to 38.8% and improve low-budget perplexity and FID versus standard masked generative models on text and images.
TCC calibrates cached representations in diffusion sampling via an offline iterative procedure that accounts for trajectory shifts, improving FID from 29.83 to 27.35 on PixArt-alpha while preserving reuse policies.
PDI-Bench computes 3D projective residuals from segmented and tracked points to quantify geometric inconsistency in AI-generated videos.
EnScale emulates high-resolution regional climate model outputs from global circulation models for multiple variables using a two-step generative process with sparse local stochastic layers and energy score optimization, including a temporally consistent variant.
MMD GANs have unbiased critic gradients but biased generator gradients from sample-based learning, and the Kernel Inception Distance provides a practical new measure for GAN convergence and dynamic learning rate adaptation.
MaMe is a differentiable matrix-only token merging method that doubles ViT-B throughput with a 2% accuracy drop on pre-trained models and enables faster, higher-quality image synthesis when paired with MaRe.
Weight-shared looped transformers trained with intra-loop self-distillation match MaskGIT-class FID/FVD at roughly 4x fewer parameters and support any-time inference across loop counts.
SDXL improves upon prior Stable Diffusion versions through a larger UNet backbone, dual text encoders, novel conditioning, and a refinement model, producing higher-fidelity images competitive with black-box state-of-the-art generators.
VideoGPT generates competitive natural videos by learning discrete latents with VQ-VAE and modeling them autoregressively with a transformer.
A VAE-based latent diffusion model trained on incomplete data maintains sample quality and imputation performance up to 50% missingness while pixel-space diffusion degrades.
A physics-constrained cGAN is trained as an image-to-image translator on remote-sensing layers to recover spatial sensitivities of urban land-use change to macroeconomic indicators via backpropagation gradients.
Experiments on real industrial time series show that partial model sharing improves diffusion model performance in bandwidth-limited non-IID settings, while full sharing stabilizes GAN training but offers less robustness than VAE or DDPM alternatives.
A GAN guided by a surrogate membrane-factor predictor generates 64-by-64 height-map shell shapes, but the physical-constraint implementation and independent validation are insufficient to establish compression-only funicular shells.
A responsible computing framework substitutes real protest imagery with labeled synthetic reproductions from conditional image synthesis to enable privacy-aware analysis of collective action patterns.
X-VAE uses empirical statistics from a pretrained autoencoder to set a data-adaptive Gaussian prior and introduces a latent scaling factor for controllable generation.
A conditional Wasserstein GAN generates plausible future SWI drought trajectories for French insurance risk management under climate change.
Empirical measurement of adversarial example transferability between VGG and Inception model classes with methodological refinements to attack strength selection, perturbation clipping, and evaluation via SSIM.
citing papers explorer
-
AGAN: Towards Automated Design of Generative Adversarial Networks
AGAN is the first neural architecture search method for GANs that discovers architectures outperforming state-of-the-art on CIFAR-10 unsupervised image generation and competitive on supervised tasks.
-
Setting-Matched and Semantics-Scaled Benchmarking of One-Step Generative Models Against Multistep Diffusion and Flow Models
Matched benchmarking reveals FID misleads in few-step regimes under CFG, prompting CLIP-scaled and PickScore-scaled FID and IS variants for better semantic evaluation of one-step image generators.
-
Omni2Sound: Towards Unified Video-Text-to-Audio Generation
A single DiT-based diffusion model unifies video-to-audio, text-to-audio, and joint video-text-to-audio generation, supported by a new 470k-pair dataset and three-stage progressive training that resolves task competition.
-
Diffusion Models Beat GANs on Image Synthesis
Diffusion models with architecture improvements and classifier guidance achieve superior FID scores to GANs on unconditional and conditional ImageNet image synthesis.
-
Fixed-Point Masked Generative Modeling
FP-MGMs with consistency loss and three-state reuse (CoFRe) reduce parameters by up to 38.8% and improve low-budget perplexity and FID versus standard masked generative models on text and images.
-
Trajectory-Consistent Calibration for Cache-Accelerated Diffusion Models
TCC calibrates cached representations in diffusion sampling via an offline iterative procedure that accounts for trajectory shifts, improving FID from 29.83 to 27.35 on PixArt-alpha while preserving reuse policies.
-
Quantitative Video World Model Evaluation for Geometric-Consistency
PDI-Bench computes 3D projective residuals from segmented and tracked points to quantify geometric inconsistency in AI-generated videos.
-
EnScale: Temporally-consistent multivariate generative downscaling via proper scoring rules
EnScale emulates high-resolution regional climate model outputs from global circulation models for multiple variables using a two-step generative process with sparse local stochastic layers and energy score optimization, including a temporally consistent variant.
-
Demystifying MMD GANs
MMD GANs have unbiased critic gradients but biased generator gradients from sample-based learning, and the Kernel Inception Distance provides a practical new measure for GAN convergence and dynamic learning rate adaptation.
-
MaMe & MaRe: Matrix-Based Token Merging and Restoration for Efficient Visual Perception and Synthesis
MaMe is a differentiable matrix-only token merging method that doubles ViT-B throughput with a 2% accuracy drop on pre-trained models and enables faster, higher-quality image synthesis when paired with MaRe.
-
ELT: Elastic Looped Transformers for Visual Generation
Weight-shared looped transformers trained with intra-loop self-distillation match MaskGIT-class FID/FVD at roughly 4x fewer parameters and support any-time inference across loop counts.
-
SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis
SDXL improves upon prior Stable Diffusion versions through a larger UNet backbone, dual text encoders, novel conditioning, and a refinement model, producing higher-fidelity images competitive with black-box state-of-the-art generators.
-
VideoGPT: Video Generation using VQ-VAE and Transformers
VideoGPT generates competitive natural videos by learning discrete latents with VQ-VAE and modeling them autoregressively with a transformer.
-
Latent Diffusion for Missing Data
A VAE-based latent diffusion model trained on incomplete data maintains sample quality and imputation performance up to 50% missingness while pixel-space diffusion degrades.
-
Spatial sensitivity analysis for urban land use prediction with physics-constrained conditional generative adversarial networks
A physics-constrained cGAN is trained as an image-to-image translator on remote-sensing layers to recover spatial sensitivities of urban land-use change to macroeconomic indicators via backpropagation gradients.
-
On the Tradeoffs of On-Device Generative Models in Federated Predictive Maintenance Systems
Experiments on real industrial time series show that partial model sharing improves diffusion model performance in bandwidth-limited non-IID settings, while full sharing stabilizes GAN training but offers less robustness than VAE or DDPM alternatives.
-
Physics-informed, Generative Adversarial Design of Funicular Shells
A GAN guided by a surrogate membrane-factor predictor generates 64-by-64 height-map shell shapes, but the physical-constraint implementation and independent validation are insufficient to establish compression-only funicular shells.
-
Protecting and Preserving Protest Dynamics for Responsible Analysis
A responsible computing framework substitutes real protest imagery with labeled synthetic reproductions from conditional image synthesis to enable privacy-aware analysis of collective action patterns.
-
eXact-Prior Variational Autoencoder (X-VAE): Learning Data-Adaptive Gaussian Mixture Priors for Latent Distributions
X-VAE uses empirical statistics from a pretrained autoencoder to set a data-adaptive Gaussian prior and introduces a latent scaling factor for controllable generation.
-
A Wasserstein GAN-based climate scenario generator for risk management and insurance: the case of soil subsidence
A conditional Wasserstein GAN generates plausible future SWI drought trajectories for French insurance risk management under climate change.
-
Measuring the Transferability of Adversarial Examples
Empirical measurement of adversarial example transferability between VGG and Inception model classes with methodological refinements to attack strength selection, perturbation clipping, and evaluation via SSIM.