Large Scale Adversarial Representation Learning

Abstract

Adversarially trained generative models (GANs) have recently achieved compelling image synthesis results. But despite early successes in using GANs for unsupervised representation learning, they have since been superseded by approaches based on self-supervision. In this work we show that progress in image generation quality translates to substantially improved representation learning performance. Our approach, BigBiGAN, builds upon the state-of-the-art BigGAN model, extending it to representation learning by adding an encoder and modifying the discriminator. We extensively evaluate the representation learning and generation capabilities of these BigBiGAN models, demonstrating that these generation-based models achieve the state of the art in unsupervised representation learning on ImageNet, as well as in unconditional image generation. Pretrained BigBiGAN models -- including image generators and encoders -- are available on TensorFlow Hub (https://tfhub.dev/s?publisher=deepmind&q=bigbigan).

arXiv comment32 pages. In proceedings of NeurIPS 2019. This is the camera-ready version of the paper, with supplementary material included as appendices

References in corpus (13)

• Auto-Encoding Variational Bayes
• Representation Learning with Contrastive Predictive Coding
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Spectral Normalization for Generative Adversarial Networks
• Neural Discrete Representation Learning
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• Variational Approaches for Auto-Encoding Generative Adversarial Networks
• The Reversible Residual Network: Backpropagation Without Storing Activations
• Understanding and Improving Convolutional Neural Networks via Concatenated Rectified Linear Units
• High-Fidelity Image Generation With Fewer Labels
• Geometric GAN
• TF-Replicator: Distributed Machine Learning for Researchers

Cited by in corpus (87)

• A Simple Framework for Contrastive Learning of Visual Representations
• Unsupervised Learning of Visual Features by Contrasting Cluster Assignments
• Momentum Contrast for Unsupervised Visual Representation Learning
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• What Makes for Good Views for Contrastive Learning?
• Contrastive Multiview Coding
• Prototypical Contrastive Learning of Unsupervised Representations
• Big Self-Supervised Models are Strong Semi-Supervised Learners
• MetNet: A Neural Weather Model for Precipitation Forecasting
• Adversarial Text-to-Image Synthesis: A Review
• Momentum Contrastive Learning for Few-Shot COVID-19 Diagnosis from Chest CT Images
• Image Representations Learned With Unsupervised Pre-Training Contain Human-like Biases
• Learning Representation for Clustering via Prototype Scattering and Positive Sampling
• The Origins and Prevalence of Texture Bias in Convolutional Neural Networks
• StarGAN v2: Diverse Image Synthesis for Multiple Domains
• Vector-quantized Image Modeling with Improved VQGAN
• Efficient Self-supervised Vision Transformers for Representation Learning
• Reconstructing Natural Scenes from fMRI Patterns using BigBiGAN
• A Good Image Generator Is What You Need for High-Resolution Video Synthesis
• A Survey of Deep Learning for Scientific Discovery
• LOGAN: Latent Optimisation for Generative Adversarial Networks
• Whitening for Self-Supervised Representation Learning
• Contrastive Learning of Single-Cell Phenotypic Representations for Treatment Classification
• Contrastive Learning with Stronger Augmentations
• Support-set bottlenecks for video-text representation learning
• Solving Inverse Problems by Joint Posterior Maximization with Autoencoding Prior
• Using latent space regression to analyze and leverage compositionality in GANs
• Cross-Camera Feature Prediction for Intra-Camera Supervised Person Re-identification across Distant Scenes
• A Comprehensive Review and a Taxonomy of Edge Machine Learning: Requirements, Paradigms, and Techniques
• Viewmaker Networks: Learning Views for Unsupervised Representation Learning
• Towards the Generalization of Contrastive Self-Supervised Learning
• Understanding and Mitigating Exploding Inverses in Invertible Neural Networks
• Understanding Negative Samples in Instance Discriminative Self-supervised Representation Learning
• Unsupervised Feature Learning by Cross-Level Instance-Group Discrimination
• Transformation-based Adversarial Video Prediction on Large-Scale Data
• Image Morphing with Perceptual Constraints and STN Alignment
• Data clustering: a fundamental method in data science and management
• Joint Contrastive Learning with Infinite Possibilities
• Distilling Localization for Self-Supervised Representation Learning
• Spatially Consistent Representation Learning
• MotionVideoGAN: A Novel Video Generator Based on the Motion Space Learned from Image Pairs
• Towards Domain-Agnostic Contrastive Learning
• Triple Generative Adversarial Networks
• AlphaGAN: Fully Differentiable Architecture Search for Generative Adversarial Networks
• On-manifold Adversarial Data Augmentation Improves Uncertainty Calibration
• CSNNs: Unsupervised, Backpropagation-free Convolutional Neural Networks for Representation Learning
• Self-Supervision, Remote Sensing and Abstraction: Representation Learning Across 3 Million Locations
• For self-supervised learning, Rationality implies generalization, provably
• Understanding and Stabilizing GANs' Training Dynamics with Control Theory
• Reachability Embeddings: Scalable Self-Supervised Representation Learning from Mobility Trajectories for Multimodal Geospatial Computer Vision
• Scalable Self-Supervised Representation Learning from Spatiotemporal Motion Trajectories for Multimodal Computer Vision
• 4D Semantic Cardiac Magnetic Resonance Image Synthesis on XCAT Anatomical Model
• Finding online neural update rules by learning to remember
• Generative Models as a Data Source for Multiview Representation Learning
• MixNMatch: Multifactor Disentanglement and Encoding for Conditional Image Generation
• Bidirectional Generative Modeling Using Adversarial Gradient Estimation
• Distribution-induced Bidirectional Generative Adversarial Network for Graph Representation Learning
• Generalized Energy Based Models
• Generative Hierarchical Features from Synthesizing Images
• Can Pretext-Based Self-Supervised Learning Be Boosted by Downstream Data? A Theoretical Analysis
• Low Distortion Block-Resampling with Spatially Stochastic Networks
• Improved Transformer for High-Resolution GANs
• Training Generative Adversarial Networks in One Stage
• Jigsaw Clustering for Unsupervised Visual Representation Learning
• Clustering-Based Representation Learning through Output Translation and Its Application to Remote--Sensing Images
• DEff-GAN: Diverse Attribute Transfer for Few-Shot Image Synthesis
• Alleviation of Gradient Exploding in GANs: Fake Can Be Real
• Don't miss the Mismatch: Investigating the Objective Function Mismatch for Unsupervised Representation Learning
• D-square-B: Deep Distribution Bound for Natural-looking Adversarial Attack
• DyStyle: Dynamic Neural Network for Multi-Attribute-Conditioned Style Editing
• Algorithm Fairness in AI for Medicine and Healthcare
• Lessons Learned from the Training of GANs on Artificial Datasets
• Exploiting Spatial Dimensions of Latent in GAN for Real-time Image Editing
• Low-Rank Subspaces in GANs
• Temporal Knowledge Consistency for Unsupervised Visual Representation Learning
• Improving Image Clustering through Sample Ranking and Its Application to remote--sensing images
• Exploiting Chain Rule and Bayes' Theorem to Compare Probability Distributions
• IID-GAN: an IID Sampling Perspective for Regularizing Mode Collapse
• IGAN: Inferent and Generative Adversarial Networks
• OBoW: Online Bag-of-Visual-Words Generation for Self-Supervised Learning
• Learning to infer in recurrent biological networks
• Self-Supervised Learning for Large-Scale Unsupervised Image Clustering
• Illiterate DALL-E Learns to Compose
• OmiTrans: generative adversarial networks based omics-to-omics translation framework
• MC-SSL0.0: Towards Multi-Concept Self-Supervised Learning
• Deep Automodulators
• Natural Image Manipulation for Autoregressive Models Using Fisher Scores