cGANs with Projection Discriminator

2018/02/15 by Takeru Miyato, Miyato, Takeru, Masanori Koyama +1 · 26 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · Engineering · Mathematics · #Cell Image Analysis Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Image Processing Techniques and Applications #Machine Learning (cs.LG) #Machine Learning (stat.ML) #cs.CV #cs.LG #stat.ML

paper · pdf · doi:10.48550/arxiv.1802.05637

Published as a conference paper at ICLR 2018

openalex publication_date 2018/02/15 · arxiv created 2018/08/15 · arxiv updated 2018/08/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We propose a novel, projection based way to incorporate the conditional information into the discriminator of GANs that respects the role of the conditional information in the underlining probabilistic model. This approach is in contrast with most frameworks of conditional GANs used in application today, which use the conditional information by concatenating the (embedded) conditional vector to the feature vectors. With this modification, we were able to significantly improve the quality of the class conditional image generation on ILSVRC2012 (ImageNet) 1000-class image dataset from the current state-of-the-art result, and we achieved this with a single pair of a discriminator and a generator. We were also able to extend the application to super-resolution and succeeded in producing highly discriminative super-resolution images. This new structure also enabled high quality category transformation based on parametric functional transformation of conditional batch normalization layers in the generator.

Citations

• ImageNet Large Scale Visual Recognition Challenge
• Improved Techniques for Training GANs
• Conditional Image Synthesis With Auxiliary Classifier GANs
• Invertible Conditional GANs for image editing
• Improved Training of Wasserstein GANs
• Semi-supervised Conditional GANs
• Hierarchical Implicit Models and Likelihood-Free Variational Inference
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Cited by

• Adversarial Flow Models
• Improving conditional generative adversarial networks for inverse design of plasmonic structures
• Contrastive Diffusion Alignment: Learning Structured Latents for Controllable Generation
• Robust Learning of Diffusion Models with Extremely Noisy Conditions
• SONA: Learning Conditional, Unconditional, and Mismatching-Aware Discriminator
• Learning Robust Diffusion Models from Imprecise Supervision
• Conditional Pseudo-Supervised Contrast for Data-Free Knowledge Distillation
• One-shot Conditional Sampling: MMD meets Nearest Neighbors
• Scalable GANs with Transformers
• Shilling Recommender Systems by Generating Side-feature-aware Fake User Profiles
• ArtifactGen: Benchmarking WGAN-GP vs Diffusion for Label-Aware EEG Artifact Synthesis
• Conditional Fetal Brain Atlas Learning for Automatic Tissue Segmentation
• GeMix: Conditional GAN-Based Mixup for Improved Medical Image Augmentation
• Super-resolution of turbulent velocity fields in two-way coupled particle-laden flows
• Learning Spatial Pyramid Attentive Pooling in Image Synthesis and Image-to-Image Translation
• Robustness of Conditional GANs to Noisy Labels
• MarioNETte: Few-shot Face Reenactment Preserving Identity of Unseen Targets
• Identifying and Exploiting Structures for Reliable Deep Learning
• MixDefense: A Defense-in-Depth Framework for Adversarial Example Detection Based on Statistical and Semantic Analysis
• Generative Adversarial Learning via Kernel Density Discrimination
• Disentangling Latent Space for VAE by Label Relevant/Irrelevant Dimensions
• On Computation and Generalization of GANs with Spectrum Control
• Continuous Conditional Generative Adversarial Networks (cGAN) with Generator Regularization
• PConv: simple yet effective convolutional layer for generative adversarial network
• Towards Amortized Ranking-Critical Training for Collaborative Filtering
• A Comprehensive Survey of Synthetic Tabular Data Generation

Related

• Are conditional GANs explicitly conditional?
• cGANs with Conditional Convolution Layer
• Dual Projection Generative Adversarial Networks for Conditional Image Generation
• cGANs with Multi-Hinge Loss
• A Unified View of cGANs with and without Classifiers