Large Scale GAN Training for High Fidelity Natural Image Synthesis

Despite recent progress in generative image modeling, successfully generating high-resolution, diverse samples from complex datasets such as ImageNet remains an elusive goal. To this end, we train Generative Adversarial Networks at the largest scale yet attempted, and study the instabilities specific to such scale. We find that applying orthogonal regularization to the generator renders it amenable to a simple "truncation trick," allowing fine control over the trade-off between sample fidelity and variety by reducing the variance of the Generator's input. Our modifications lead to models which set the new state of the art in class-conditional image synthesis. When trained on ImageNet at 128x128 resolution, our models (BigGANs) achieve an Inception Score (IS) of 166.5 and Frechet Inception Distance (FID) of 7.4, improving over the previous best IS of 52.52 and FID of 18.6.

• Publication year

  2018

• Venue

  International Conference on Learning Representations

• Publication date

  2018-09-27

• Fields of study

  Mathematics, Computer Science

• Identifiers
  arXiv 1809.11096
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Improving GANs Using Optimal Transport

  2018cited by this paper
• Demystifying MMD GANs

  2018cited by this paper
• A Note on the Inception Score

  2018cited by this paper
• Is Generator Conditioning Causally Related to GAN Performance?

  2018cited by this paper
• cGANs with Projection Discriminator

  2018influential reference
• On Convergence and Stability of GANs

  2018cited by this paper
• The Unusual Effectiveness of Averaging in GAN Training

  2018cited by this paper
• GENERATIVE ADVERSARIAL NETS

  2018influential reference
• Comparing Generative Adversarial Network Techniques for Image Creation and Modification

  2018cited by this paper
• Which Training Methods for GANs do actually Converge?

  2018influential reference
• Spectral Normalization for Generative Adversarial Networks

  2018influential reference
• Self-Attention Generative Adversarial Networks

  2018influential reference
• GANs Trained by a Two Time-Scale Update Rule Converge to a Nash Equilibrium

  2017cited by this paper
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium

  2017cited by this paper
• Wasserstein Generative Adversarial Networks

  2017cited by this paper
• Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step

  2017cited by this paper
• Progressive Growing of GANs for Improved Quality, Stability, and Variation

  2017influential reference
• Non-local Neural Networks

  2017cited by this paper
• Revisiting Unreasonable Effectiveness of Data in Deep Learning Era

  2017influential reference
• Modulating early visual processing by language

  2017cited by this paper
• Geometric GAN

  2017cited by this paper
• The Cramer Distance as a Solution to Biased Wasserstein Gradients

  2017cited by this paper
• Improved Training of Wasserstein GANs

  2017influential reference
• Hierarchical Implicit Models and Likelihood-Free Variational Inference

  2017cited by this paper
• FiLM: Visual Reasoning with a General Conditioning Layer

  2017cited by this paper
• Megapixel Size Image Creation using Generative Adversarial Networks

  2017cited by this paper
• Improved Techniques for Training GANs

  2016influential reference
• Neural Photo Editing with Introspective Adversarial Networks

  2016influential reference
• Deconvolution and Checkerboard Artifacts

  2016cited by this paper
• TensorFlow: A system for large-scale machine learning

  2016cited by this paper
• Amortised MAP Inference for Image Super-resolution

  2016cited by this paper
• Least Squares Generative Adversarial Networks

  2016cited by this paper
• f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization

  2016cited by this paper
• InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets

  2016cited by this paper
• A Learned Representation For Artistic Style

  2016cited by this paper
• Conditional Image Synthesis with Auxiliary Classifier GANs

  2016cited by this paper
• Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks

  2016influential reference
• On the Quantitative Analysis of Decoder-Based Generative Models

  2016cited by this paper
• Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks

  2015cited by this paper
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

  2015influential reference
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

  2015influential reference
• Deep Residual Learning for Image Recognition

  2015influential reference
• A note on the evaluation of generative models

  2015cited by this paper
• Rethinking the Inception Architecture for Computer Vision

  2015cited by this paper
• Adam: A Method for Stochastic Optimization

  2014cited by this paper
• Dropout: a simple way to prevent neural networks from overfitting

  2014cited by this paper
• Conditional Generative Adversarial Nets

  2014influential reference
• Very Deep Convolutional Networks for Large-Scale Image Recognition

  2014cited by this paper
• ImageNet Large Scale Visual Recognition Challenge

  2014cited by this paper
• Exact solutions to the nonlinear dynamics of learning in deep linear neural networks

  2013influential reference
• Understanding the difficulty of training deep feedforward neural networks

  2010cited by this paper
• Learning Multiple Layers of Features from Tiny Images

  2009influential reference
• Eigenvalue computation in the 20th century

  2000cited by this paper

• Duality Models: An Embarrassingly Simple One-step Generation Paradigm

  2026cites this paper
• SnapGen++: Unleashing Diffusion Transformers for Efficient High-Fidelity Image Generation on Edge Devices

  2026cites this paper
• Consistency-Regularized GAN for Few-Shot SAR Target Recognition

  2026cites this paper
• Physics Encoded Spatial and Temporal Generative Adversarial Network for Tropical Cyclone Image Super-resolution

  2026cites this paper
• FlowConsist: Make Your Flow Consistent with Real Trajectory

  2026cites this paper
• Forward-Guided and Reverse-Resampling Diffusion Model for GPR B-Scan Data Restoration and Generation

  2026cites this paper
• Diffusion Epistemic Uncertainty with Asymmetric Learning for Diffusion-Generated Image Detection

  2026cites this paper
• Mirai: Autoregressive Visual Generation Needs Foresight

  2026cites this paper
• Beyond Cropping and Rotation: Automated Evolution of Powerful Task-Specific Augmentations with Generative Models

  2026cites this paper
• Foundation Models for Medical Imaging: Status, Challenges, and Directions

  2026cites this paper
• A Proper Scoring Rule for Virtual Staining

  2026cites this paper
• Flexible Inspection of Defects in Liquid Crystal Display Panels: A Review

  2026cites this paper
• One-step Latent-free Image Generation with Pixel Mean Flows

  2026cites this paper
• Inverse Multi-Objective Design of Three-Dimensional Plate-Based Heterogeneous Mechanical Metamaterials

  2026cites this paper
• Enhancing Few-Shot Surface Defect Recognition via Pre-Trained Large Generative Models

  2026cites this paper
• Generalizable and Adaptive Continual Learning Framework for AI-generated Image Detection

  2026cites this paper
• LatentZoom: Seamless Scaling in Generative Latent Space for Visual Exploration of Local Performance in Deep Neural Networks

  2026cites this paper
• Soft Tail-dropping for Adaptive Visual Tokenization

  2026cites this paper
• Exploring the transformer-based and diffusion-based models for single image deblurring

  2026cites this paper
• Over-Alignment vs Over-Fitting: The Role of Feature Learning Strength in Generalization

  2026influential citation
• Dual-End Consistency Model

  2026cites this paper
• GICDM: Mitigating Hubness for Reliable Distance-Based Generative Model Evaluation

  2026cites this paper
• SAPGAN: Style-Attentive Progressive Growing Generative Adversarial Network for Limited-Data Image Synthesis

  2026cites this paper
• Diversity over Uniformity: Rethinking Representation in Generated Image Detection

  2026cites this paper
• Controlled Face Manipulation and Synthesis for Data Augmentation

  2026cites this paper
• Latent Forcing: Reordering the Diffusion Trajectory for Pixel-Space Image Generation

  2026cites this paper
• Generative Modeling via Drifting

  2026cites this paper
• Creative Image Generation with Diffusion Model

  2026cites this paper
• Ultraman: ultra-fast and high-resolution texture generation for 3D human reconstruction from a single image

  2026cites this paper
• DSA-Diff: Dynamic schedule alignment for training-Inference consistent modality translation in x-prediction diffusion model.

  2026cites this paper
• Augmentation of 3D virtual aggregate database using deep convolutional Wasserstein generative adversarial networks

  2026cites this paper
• Potential of artificial intelligence in deepfake media: From generation to detection mechanisms, state-of-the-art, and challenges

  2026cites this paper
• A Hybrid Architecture Combining Physical Modeling and Neural Networks for Piano Sound Synthesis

  2026cites this paper
• Bridging the Discrete-Continuous Gap: Unified Multimodal Generation via Coupled Manifold Discrete Absorbing Diffusion

  2026cites this paper
• Water2LandNet: generative adversarial networks for UAV image dewatering

  2026cites this paper
• Deepfake detection using a distinctive eye signature and the entropy heat map of the image texture

  2026cites this paper
• Synthetic Packet Traffic Generative Adversarial Networks in Multi Agents With Peer-to-Peer and Global Priority Queue Generation

  2026cites this paper
• Bone-conduction Guided Multimodal Speech Enhancement with Conditional Diffusion Models

  2026cites this paper
• An analytical review of GANs: technical evolution, architectures, applications, datasets, and challenges

  2026cites this paper
• An Improved Diffusion Model for Generating Images of a Single Category of Food on a Small Dataset

  2026cites this paper
• Development of global diffusion filling model for voids filling of Digital elevation models (DEMs)

  2026cites this paper
• Your AI-Generated Image Detector Can Secretly Achieve SOTA Accuracy, If Calibrated

  2026cites this paper
• Temporal Pair Consistency for Variance-Reduced Flow Matching

  2026influential citation
• Dual-Manifold Gradient Rectification: Reconciling Topology and Texture in Industrial Anomaly Synthesis

  2026cites this paper
• Why retrieve when you can edit: A fast conditional StyleGAN latent editing method

  2026cites this paper
• Latent Regularization in Generative Test Input Generation

  2026cites this paper
• TranX-Adapter: Bridging Artifacts and Semantics within MLLMs for Robust AI-generated Image Detection

  2026cites this paper
• Optimization and cross model validation of femtosecond laser processed nickel surface structural parameters based on deep learning

  2026cites this paper
• Physical Evaluation of Naturalistic Adversarial Patches for Camera-Based Traffic-Sign Detection

  2026cites this paper
• Error as Signal: Stiffness-Aware Diffusion Sampling via Embedded Runge-Kutta Guidance

  2026cites this paper
• Revolutionizing sentiment analysis with generative AI: techniques, trends, and challenges

  2026cites this paper
• SimLBR: Learning to Detect Fake Images by Learning to Detect Real Images

  2026cites this paper
• Exposing Diversity Bias in Deep Generative Models: Statistical Origins and Correction of Diversity Error

  2026cites this paper
• Image Generation with a Sphere Encoder

  2026cites this paper
• Deep Generative Models for Node Embedding and Neighborhood Prediction in Dynamic Graphs of Recommendation Systems

  2026cites this paper
• FedPDM: Representation enhanced federated learning with privacy preserving diffusion models

  2026cites this paper
• Dispelling the Curse of Singularities in Neural Network Optimizations

  2026cites this paper
• Color Matters: Demosaicing-Guided Color Correlation Training for Generalizable AI-Generated Image Detection

  2026cites this paper
• DREAMSTATE: Diffusing States and Parameters for Recurrent Large Language Models

  2026cites this paper
• RealStats: A Rigorous Real-Only Statistical Framework for Fake Image Detection

  2026cites this paper
• Few-shot GAN adaptation for high-fidelity and diverse crack image generation in dam damage detection

  2026cites this paper
• S^2F-Net:A Robust Spatial-Spectral Fusion Framework for Cross-Model AIGC Detection

  2026cites this paper
• Gradient-Guided Diffusion-Based Restoration of Extremely Compressed Backgrounds for Video Coding for Machines

  2026cites this paper
• BDTest: A Diversity-Oriented Test Case Generation Framework for Deep Neural Networks in 6G-IOT

  2026cites this paper
• Normalized clipping: A privacy-enhanced method in differentially private GANs

  2026cites this paper
• Spatio-Temporal Diffusion Model for Cellular Traffic Generation

  2026cites this paper
• CoLeQ: Improving Data-Free Quantization via Contrastive Learning

  2026cites this paper
• Unknown Aware AI-Generated Content Attribution

  2026cites this paper
• Stroke Outcome and Evolution Prediction from CT Brain Using a Spatiotemporal Diffusion Autoencoder

  2026cites this paper
• AI-generated image detection algorithm based on classical-quantum hybrid neural network

  2026cites this paper
• Leveraging 3D Representation Alignment and RGB Pretrained Priors for LiDAR Scene Generation

  2026cites this paper
• From Easy to Hard++: Promoting Differentially Private Image Synthesis Through Spatial-Frequency Curriculum

  2026cites this paper
• Improving StyleGAN-ADA with efficient channel attention and multidimensional enhancement heuristic for solder joint defect detection

  2026cites this paper
• De Novo Design of Large Polypeptides Using a Lightweight Diffusion Model Integrating LSTM and Attention Mechanism Under Per-Residue Secondary Structure Constraints

  2025cites this paper
• Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generation

  2025cites this paper
• Towards Extensible Detection of AI-Generated Images via Content-Agnostic Adapter-Based Category-Aware Incremental Learning

  2025cites this paper
• UIFace: Unleashing Inherent Model Capabilities to Enhance Intra-Class Diversity in Synthetic Face Recognition

  2025cites this paper
• A Discrete Index Graph Diffusion Model for 3D Meshes Synthesis

  2025cites this paper
• FlexVAR: Flexible Visual Autoregressive Modeling without Residual Prediction

  2025cites this paper
• KeyBoxGAN: enhancing 2D object detection through annotated and editable image synthesis

  2025cites this paper
• Reducing the Content Bias for AI-generated Image Detection

  2025cites this paper
• Autoregressive Image Generation with Vision Full-view Prompt

  2025cites this paper
• Fractal Generative Models

  2025cites this paper
• Methods and trends in detecting AI-generated images: A comprehensive review

  2025cites this paper
• One-step Diffusion Models with f-Divergence Distribution Matching

  2025cites this paper
• Generative artificial intelligence: a historical perspective

  2025cites this paper
• Optical imaging combined with artificial intelligence in plant disease detection: a comprehensive review

  2025cites this paper
• LAST: Utilizing Synthetic Image Style Transfer to Tackle Domain Shift in Aerial Image Segmentation

  2025cites this paper
• Diffusion Models without Classifier-free Guidance

  2025cites this paper
• Beyond Known Fakes: Generalized Detection of AI-Generated Images via Post-hoc Distribution Alignment

  2025cites this paper
• Characterizing Photorealism and Artifacts in Diffusion Model-Generated Images

  2025cites this paper
• DiffEx: Explaining a Classifier with Diffusion Models to Identify Microscopic Cellular Variations

  2025cites this paper
• A diffusion model-based dual domain approach for CT metal artifact reduction

  2025cites this paper
• The Vendiscope: An Algorithmic Microscope For Data Collections

  2025cites this paper
• Personalized Image Generation with Deep Generative Models: A Decade Survey

  2025cites this paper
• Score-of-Mixture Training: One-Step Generative Model Training Made Simple via Score Estimation of Mixture Distributions

  2025cites this paper
• On the caveats of AI autophagy

  2025cites this paper
• Spread them Apart: Towards Robust Watermarking of Generated Content

  2025cites this paper
• Machine intelligence for interpretation and preservation of built heritage

  2025cites this paper
• Make the Fastest Faster: Importance Mask Synthesis for Interactive Volume Visualization Using Reconstruction Neural Networks

  2025cites this paper