Adversarial Feature Learning

Abstract

The ability of the Generative Adversarial Networks (GANs) framework to learn generative models mapping from simple latent distributions to arbitrarily complex data distributions has been demonstrated empirically, with compelling results showing that the latent space of such generators captures semantic variation in the data distribution. Intuitively, models trained to predict these semantic latent representations given data may serve as useful feature representations for auxiliary problems where semantics are relevant. However, in their existing form, GANs have no means of learning the inverse mapping -- projecting data back into the latent space. We propose Bidirectional Generative Adversarial Networks (BiGANs) as a means of learning this inverse mapping, and demonstrate that the resulting learned feature representation is useful for auxiliary supervised discrimination tasks, competitive with contemporary approaches to unsupervised and self-supervised feature learning.

Published as a conference paper at ICLR 2017. Changelog: (v7) Table 2 results improved 1-2% due to averaging predictions over 10 crops at test time, as done in Noroozi & Favaro; Table 3 VOC classification results slightly improved due to minor bugfix. (See v6 changelog for previous versions.)

References in corpus (5)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Sequence to Sequence Learning with Neural Networks
• Caffe: Convolutional Architecture for Fast Feature Embedding
• Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
• Adversarially Learned Inference

Cited by in corpus (336)

• Bootstrap your own latent: A new approach to self-supervised Learning
• Generative Adversarial Network in Medical Imaging: A Review
• Unsupervised Representation Learning by Predicting Image Rotations
• NIPS 2016 Tutorial: Generative Adversarial Networks
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• Active Learning for Convolutional Neural Networks: A Core-Set Approach
• Adversarially Learned Inference
• CyCADA: Cycle-Consistent Adversarial Domain Adaptation
• Invertible Conditional GANs for image editing
• VEEGAN: Reducing Mode Collapse in GANs using Implicit Variational Learning
• Deep Clustering for Unsupervised Learning of Visual Features
• Deep Generative Design: Integration of Topology Optimization and Generative Models
• Deep Autoencoding Models for Unsupervised Anomaly Segmentation in Brain MR Images
• How Generative Adversarial Networks and Their Variants Work: An Overview
• Recent Advances in Autoencoder-Based Representation Learning
• State Representation Learning for Control: An Overview
• Revisiting Unreasonable Effectiveness of Data in Deep Learning Era
• Multimodal Unsupervised Image-to-Image Translation
• Generative Adversarial Networks in Computer Vision: A Survey and Taxonomy
• Self-supervised learning methods and applications in medical imaging analysis: A survey
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• X-ModalNet: A Semi-Supervised Deep Cross-Modal Network for Classification of Remote Sensing Data
• BAGAN: Data Augmentation with Balancing GAN
• Variational Approaches for Auto-Encoding Generative Adversarial Networks
• Neural Photo Editing with Introspective Adversarial Networks
• An Overview of Deep Semi-Supervised Learning
• Mode Regularized Generative Adversarial Networks
• Good Semi-supervised Learning that Requires a Bad GAN
• Triple Generative Adversarial Nets
• Optimizing the Latent Space of Generative Networks
• Learning a Driving Simulator
• Self-supervised Visual Feature Learning with Deep Neural Networks: A Survey
• The Numerics of GANs
• Unsupervised Feature Learning via Non-Parametric Instance-level Discrimination
• An Introduction to Image Synthesis with Generative Adversarial Nets
• Unsupervised Anomaly Detection with Generative Adversarial Networks to Guide Marker Discovery
• Generative Adversarial Networks (GANs) in Networking: A Comprehensive Survey & Evaluation
• Do GANs actually learn the distribution? An empirical study
• Deep Generative Dual Memory Network for Continual Learning
• Adversarial Feature Matching for Text Generation
• Towards Automated ICD Coding Using Deep Learning
• The Origins and Prevalence of Texture Bias in Convolutional Neural Networks
• The Robust Manifold Defense: Adversarial Training using Generative Models
• Lifelong Generative Modeling
• Semi-supervised Learning with GANs: Manifold Invariance with Improved Inference
• Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step
• Pretrained Transformers as Universal Computation Engines
• Generative Models for Automatic Chemical Design
• Towards Realistic Individual Recourse and Actionable Explanations in Black-Box Decision Making Systems
• Hierarchical Implicit Models and Likelihood-Free Variational Inference
• Probabilistic Forecasting of Sensory Data with Generative Adversarial Networks - ForGAN
• Precise Recovery of Latent Vectors from Generative Adversarial Networks
• Connecting Generative Adversarial Networks and Actor-Critic Methods
• Loss is its own Reward: Self-Supervision for Reinforcement Learning
• Image-to-image translation for cross-domain disentanglement
• Meta-Learning Update Rules for Unsupervised Representation Learning
• Self-Contrastive Learning with Hard Negative Sampling for Self-supervised Point Cloud Learning
• Learning Plannable Representations with Causal InfoGAN
• Unsupervised Embedding Learning via Invariant and Spreading Instance Feature
• Local Aggregation for Unsupervised Learning of Visual Embeddings
• Triangle Generative Adversarial Networks
• Scattering Networks for Hybrid Representation Learning
• Deep Representation Learning in Speech Processing: Challenges, Recent Advances, and Future Trends
• Adversarially Regularized Graph Autoencoder for Graph Embedding
• Reconstructing Natural Scenes from fMRI Patterns using BigBiGAN
• Semantic-aware Grad-GAN for Virtual-to-Real Urban Scene Adaption
• FrankenGAN: Guided Detail Synthesis for Building Mass-Models Using Style-Synchonized GANs
• Implicit Maximum Likelihood Estimation
• Label Efficient Learning of Transferable Representations across Domains and Tasks
• PSIGAN: Joint probabilistic segmentation and image distribution matching for unpaired cross-modality adaptation based MRI segmentation
• Colorization as a Proxy Task for Visual Understanding
• AET vs. AED: Unsupervised Representation Learning by Auto-Encoding Transformations rather than Data
• Variational Inference of Disentangled Latent Concepts from Unlabeled Observations
• Unsupervised Heterogeneous Coupling Learning for Categorical Representation
• Learning Beyond Human Expertise with Generative Models for Dental Restorations
• Prescribed Generative Adversarial Networks
• Adversarial Examples - A Complete Characterisation of the Phenomenon
• A Spectral Approach to Gradient Estimation for Implicit Distributions
• Correction by Projection: Denoising Images with Generative Adversarial Networks
• Multi-View Image Generation from a Single-View
• Object Discovery with a Copy-Pasting GAN
• Ensembles of Generative Adversarial Networks
• Look, Listen and Learn
• Adversarially Regularized Autoencoders
• Found in Translation: Learning Robust Joint Representations by Cyclic Translations Between Modalities
• BourGAN: Generative Networks with Metric Embeddings
• Clustering-friendly Representation Learning via Instance Discrimination and Feature Decorrelation
• Contrastive Self-supervised Neural Architecture Search
• Contrastive Learning with Stronger Augmentations
• Adversarial Symmetric Variational Autoencoder
• Adversarial Training in Affective Computing and Sentiment Analysis: Recent Advances and Perspectives
• Predicting Deeper into the Future of Semantic Segmentation
• VELC: A New Variational AutoEncoder Based Model for Time Series Anomaly Detection
• Multi-view Generative Adversarial Networks
• Challenges in Disentangling Independent Factors of Variation
• Safer Classification by Synthesis
• The Information Autoencoding Family: A Lagrangian Perspective on Latent Variable Generative Models
• MGGAN: Solving Mode Collapse using Manifold Guided Training
• Quantifying Perceptual Distortion of Adversarial Examples
• Conditional Generation of Medical Images via Disentangled Adversarial Inference
• Using latent space regression to analyze and leverage compositionality in GANs
• Style2Vec: Representation Learning for Fashion Items from Style Sets
• On Training Sample Memorization: Lessons from Benchmarking Generative Modeling with a Large-scale Competition
• Improved Training of Generative Adversarial Networks Using Representative Features
• Modeling documents with Generative Adversarial Networks
• Improving Inversion and Generation Diversity in StyleGAN using a Gaussianized Latent Space
• GAN-QP: A Novel GAN Framework without Gradient Vanishing and Lipschitz Constraint
• Structured Generative Adversarial Networks
• JointGAN: Multi-Domain Joint Distribution Learning with Generative Adversarial Nets
• Comparative Study on Generative Adversarial Networks
• Latent Constraints: Learning to Generate Conditionally from Unconditional Generative Models
• Semi-supervised Image Attribute Editing using Generative Adversarial Networks
• Inverting The Generator Of A Generative Adversarial Network (II)
• DiscrimNet: Semi-Supervised Action Recognition from Videos using Generative Adversarial Networks
• Gang of GANs: Generative Adversarial Networks with Maximum Margin Ranking
• Viewmaker Networks: Learning Views for Unsupervised Representation Learning
• Mode Collapse and Regularity of Optimal Transportation Maps
• Image Clustering using an Augmented Generative Adversarial Network and Information Maximization
• Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder
• Self-supervised EEG Representation Learning for Automatic Sleep Staging
• Conditional Image-to-Image Translation
• Adversarial Message Passing For Graphical Models
• Adversarial Variational Optimization of Non-Differentiable Simulators
• An Adversarial Regularisation for Semi-Supervised Training of Structured Output Neural Networks
• Wasserstein Dependency Measure for Representation Learning
• Image-to-Image Translation: Methods and Applications
• Geometry-Consistent Generative Adversarial Networks for One-Sided Unsupervised Domain Mapping
• A Survey on Self-supervised Pre-training for Sequential Transfer Learning in Neural Networks
• CariGAN: Caricature Generation through Weakly Paired Adversarial Learning
• Adversarial Representation Active Learning
• xGEMs: Generating Examplars to Explain Black-Box Models
• Generative Adversarial Transformers
• Hierarchical Adversarially Learned Inference
• Unsupervised Learning of Video Representations via Dense Trajectory Clustering
• Efficient GAN-based method for cyber-intrusion detection
• Feature-metric Loss for Self-supervised Learning of Depth and Egomotion
• Towards Cross-Granularity Few-Shot Learning: Coarse-to-Fine Pseudo-Labeling with Visual-Semantic Meta-Embedding
• On the Effectiveness of Least Squares Generative Adversarial Networks
• A Review of Learning with Deep Generative Models from Perspective of Graphical Modeling
• O-GAN: Extremely Concise Approach for Auto-Encoding Generative Adversarial Networks
• Theoretical limitations of Encoder-Decoder GAN architectures
• Top-k Training of GANs: Improving GAN Performance by Throwing Away Bad Samples
• Learning to Sketch with Shortcut Cycle Consistency
• Deep Generative Models for Distribution-Preserving Lossy Compression
• Visual Anomaly Detection for Images: A Survey
• A step towards procedural terrain generation with GANs
• Generative Invertible Networks (GIN): Pathophysiology-Interpretable Feature Mapping and Virtual Patient Generation
• Associative Compression Networks for Representation Learning
• Seq2Seq2Sentiment: Multimodal Sequence to Sequence Models for Sentiment Analysis
• Task Specific Adversarial Cost Function
• Effect of The Latent Structure on Clustering with GANs
• Active Image Synthesis for Efficient Labeling
• Featurized Bidirectional GAN: Adversarial Defense via Adversarially Learned Semantic Inference
• Unsupervised learning of object frames by dense equivariant image labelling
• Invariant Representations from Adversarially Censored Autoencoders
• Bangla Handwritten Digit Recognition and Generation
• Learning to Write Stylized Chinese Characters by Reading a Handful of Examples
• Estimation of Individual Treatment Effect in Latent Confounder Models via Adversarial Learning
• RadialGAN: Leveraging multiple datasets to improve target-specific predictive models using Generative Adversarial Networks
• Globally Injective ReLU Networks
• Unsupervised learning for concept detection in medical images: a comparative analysis
• Gradients as Features for Deep Representation Learning
• Primal-Dual Wasserstein GAN
• Adaptive Hierarchical Dual Consistency for Semi-Supervised Left Atrium Segmentation on Cross-Domain Data
• Unsupervised Anomaly Detection with Adversarial Mirrored AutoEncoders
• Understanding the (un)interpretability of natural image distributions using generative models
• Exploring Self-Supervised Regularization for Supervised and Semi-Supervised Learning
• Variational f-divergence Minimization
• Generating Semantic Adversarial Examples via Feature Manipulation
• Variational Inference: A Unified Framework of Generative Models and Some Revelations
• SCARF: Self-Supervised Contrastive Learning using Random Feature Corruption
• Is Discriminator a Good Feature Extractor?
• Semantic Segmentation with Generative Models: Semi-Supervised Learning and Strong Out-of-Domain Generalization
• Towards Good Practices in Self-supervised Representation Learning
• An Introduction to Person Re-identification with Generative Adversarial Networks
• Brain Tumor Anomaly Detection via Latent Regularized Adversarial Network
• Metropolis-Hastings view on variational inference and adversarial training
• An Improved Self-supervised GAN via Adversarial Training
• Triple Generative Adversarial Networks
• Galaxy Image Translation with Semi-supervised Noise-reconstructed Generative Adversarial Networks
• Self-Supervised Feature Learning by Learning to Spot Artifacts
• Local Unsupervised Learning for Image Analysis
• MCL-GAN: Generative Adversarial Networks with Multiple Specialized Discriminators
• Self-Supervised Relative Depth Learning for Urban Scene Understanding
• Improving GAN Training via Binarized Representation Entropy (BRE) Regularization
• Learning latent representations across multiple data domains using Lifelong VAEGAN
• Invert and Defend: Model-based Approximate Inversion of Generative Adversarial Networks for Secure Inference
• 3D Hand Pose Estimation using Simulation and Partial-Supervision with a Shared Latent Space
• Towards a Near Universal Time Series Data Mining Tool: Introducing the Matrix Profile
• When and How Can Deep Generative Models be Inverted?
• Generative Minimization Networks: Training GANs Without Competition
• Local Label Propagation for Large-Scale Semi-Supervised Learning
• Diverse Image Synthesis from Semantic Layouts via Conditional IMLE
• on the effectiveness of generative adversarial network on anomaly detection
• Exploring Explicit Domain Supervision for Latent Space Disentanglement in Unpaired Image-to-Image Translation
• Adversarially Approximated Autoencoder for Image Generation and Manipulation
• Scattering Transform Based Image Clustering using Projection onto Orthogonal Complement
• Neural Topic Modeling with Bidirectional Adversarial Training
• Discriminator Feature-based Inference by Recycling the Discriminator of GANs
• Audio2Gestures: Generating Diverse Gestures from Speech Audio with Conditional Variational Autoencoders
• Coupled IGMM-GANs for deep multimodal anomaly detection in human mobility data
• Mix-and-Match Tuning for Self-Supervised Semantic Segmentation
• Training VAEs Under Structured Residuals
• Unsupervised MR Motion Artifact Deep Learning using Outlier-Rejecting Bootstrap Aggregation
• Adversarial Colorization Of Icons Based On Structure And Color Conditions
• Kernel Implicit Variational Inference
• Improving Consistency and Correctness of Sequence Inpainting using Semantically Guided Generative Adversarial Network
• AVT: Unsupervised Learning of Transformation Equivariant Representations by Autoencoding Variational Transformations
• Rethinking Class Relations: Absolute-relative Supervised and Unsupervised Few-shot Learning
• A Plug-in Method for Representation Factorization in Connectionist Models
• Adversarial Graph Disentanglement
• Cycle-Consistent Adversarial Learning as Approximate Bayesian Inference
• Learning Inverse Mapping by Autoencoder based Generative Adversarial Nets
• PMC-GANs: Generating Multi-Scale High-Quality Pedestrian with Multimodal Cascaded GANs
• Discriminative-Generative Representation Learning for One-Class Anomaly Detection
• Apparel-invariant Feature Learning for Apparel-changed Person Re-identification
• Acoustic Feature Learning via Deep Variational Canonical Correlation Analysis
• A Multi-Domain Feature Learning Method for Visual Place Recognition
• Active Decision Boundary Annotation with Deep Generative Models
• In&Out : Diverse Image Outpainting via GAN Inversion
• Disentanglement for Discriminative Visual Recognition
• Towards Recovery of Conditional Vectors from Conditional Generative Adversarial Networks
• Cross-Domain Self-supervised Multi-task Feature Learning using Synthetic Imagery
• SuperFront: From Low-resolution to High-resolution Frontal Face Synthesis
• To Relieve Your Headache of Training an MRF, Take AdVIL
• Unsupervised Transfer Learning with Self-Supervised Remedy
• Distribution-induced Bidirectional Generative Adversarial Network for Graph Representation Learning
• Image-to-Image Translation with Multi-Path Consistency Regularization
• Adversarial Pixel-Level Generation of Semantic Images
• Learning Disentangled Representations of Satellite Image Time Series
• Divergence Triangle for Joint Training of Generator Model, Energy-based Model, and Inference Model
• Latent Variable Modeling for Generative Concept Representations and Deep Generative Models
• Deep Self-Supervised Representation Learning for Free-Hand Sketch
• Neural Topic Modeling with Cycle-Consistent Adversarial Training
• GANs with Variational Entropy Regularizers: Applications in Mitigating the Mode-Collapse Issue
• Joint Training of Variational Auto-Encoder and Latent Energy-Based Model
• Decomposed Adversarial Learned Inference
• BEHAVE: Behavior-Aware, Intelligent and Fair Resource Management for Heterogeneous Edge-IoT Systems
• Multimodal Image-to-Image Translation via Mutual Information Estimation and Maximization
• Reciprocal Adversarial Learning via Characteristic Functions
• Mode matching in GANs through latent space learning and inversion
• Cross-Domain Adversarial Auto-Encoder
• Global versus Localized Generative Adversarial Nets
• Facial Emotion Recognition with Noisy Multi-task Annotations
• DyStyle: Dynamic Neural Network for Multi-Attribute-Conditioned Style Editing
• Cross Domain Image Generation through Latent Space Exploration with Adversarial Loss
• Multivariate-Information Adversarial Ensemble for Scalable Joint Distribution Matching
• Semi-Implicit Generative Model
• UU-Nets Connecting Discriminator and Generator for Image to Image Translation
• Exploiting GAN Internal Capacity for High-Quality Reconstruction of Natural Images
• MIM: Mutual Information Machine
• "Best-of-Many-Samples" Distribution Matching
• Set Distribution Networks: a Generative Model for Sets of Images
• Towards Learning a Self-inverse Network for Bidirectional Image-to-image Translation
• Double cycle-consistent generative adversarial network for unsupervised conditional generation
• Adversarially Learned Mixture Model
• Unsupervised Image Classification for Deep Representation Learning
• Generative Adversarial Data Programming
• AMAD: Adversarial Multiscale Anomaly Detection on High-Dimensional and Time-Evolving Categorical Data
• Neural Bayes: A Generic Parameterization Method for Unsupervised Representation Learning
• Eliminating the Blind Spot: Adapting 3D Object Detection and Monocular Depth Estimation to 360° Panoramic Imagery
• Laplacian Denoising Autoencoder
• Conditioning Trick for Training Stable GANs
• Latent Vector Recovery of Audio GANs
• S2cGAN: Semi-Supervised Training of Conditional GANs with Fewer Labels
• Exploiting Spatial Dimensions of Latent in GAN for Real-time Image Editing
• Exploring Set Similarity for Dense Self-supervised Representation Learning
• Information Theory-Guided Heuristic Progressive Multi-View Coding
• Boosting Supervised Learning Performance with Co-training
• IVE-GAN: Invariant Encoding Generative Adversarial Networks
• Hierarchical Video Generation for Complex Data
• Adversarial Binary Coding for Efficient Person Re-identification
• D2C: Diffusion-Denoising Models for Few-shot Conditional Generation
• Unsupervised Representation Adversarial Learning Network: from Reconstruction to Generation
• Non-Exhaustive Learning Using Gaussian Mixture Generative Adversarial Networks
• Generative Models for Security: Attacks, Defenses, and Opportunities
• Non-Asymptotic Error Bounds for Bidirectional GANs
• Mode Penalty Generative Adversarial Network with adapted Auto-encoder
• Self-supervised Learning with Fully Convolutional Networks
• Train a One-Million-Way Instance Classifier for Unsupervised Visual Representation Learning
• Generative Transition Mechanism to Image-to-Image Translation via Encoded Transformation
• Classification of sparsely labeled spatio-temporal data through semi-supervised adversarial learning
• Wasserstein Introspective Neural Networks
• Inverse mapping of face GANs
• Robust Conditional GAN from Uncertainty-Aware Pairwise Comparisons
• AETv2: AutoEncoding Transformations for Self-Supervised Representation Learning by Minimizing Geodesic Distances in Lie Groups
• Weakly supervised discriminative feature learning with state information for person identification
• Lie Group Auto-Encoder
• Adversarial Learning for Image Forensics Deep Matching with Atrous Convolution
• AE-StyleGAN: Improved Training of Style-Based Auto-Encoders
• Dist-GAN: An Improved GAN using Distance Constraints
• Improving Visual Recognition using Ambient Sound for Supervision
• Recent Advancements in Self-Supervised Paradigms for Visual Feature Representation
• Disentangling Latent Space for VAE by Label Relevant/Irrelevant Dimensions
• Learning to infer in recurrent biological networks
• Multiclass non-Adversarial Image Synthesis, with Application to Classification from Very Small Sample
• Super-Selfish: Self-Supervised Learning on Images with PyTorch
• Flexible Prior Distributions for Deep Generative Models
• Collaboration among Image and Object Level Features for Image Colourisation
• Representation Learning by Reconstructing Neighborhoods
• SVMax: A Feature Embedding Regularizer
• BooVAE: Boosting Approach for Continual Learning of VAE
• Multimodal Image Synthesis with Conditional Implicit Maximum Likelihood Estimation
• Self-supervised Point Set Local Descriptors for Point Cloud Registration
• Deep Learning for Wireless Communications
• Natural Image Manipulation for Autoregressive Models Using Fisher Scores
• Nested Scale Editing for Conditional Image Synthesis
• Editable Generative Adversarial Networks: Generating and Editing Faces Simultaneously
• Analyzing the benefits of communication channels between deep learning models
• Improving Stability of LS-GANs for Audio and Speech Signals
• A Classification approach towards Unsupervised Learning of Visual Representations
• ReshapeGAN: Object Reshaping by Providing A Single Reference Image
• On Latent Distributions Without Finite Mean in Generative Models
• Explorable Tone Mapping Operators
• Uncertainty-aware Self-supervised 3D Data Association
• The Implicit Metropolis-Hastings Algorithm
• Glioblastoma Multiforme Prognosis: MRI Missing Modality Generation, Segmentation and Radiogenomic Survival Prediction
• Directional GAN: A Novel Conditioning Strategy for Generative Networks
• Contrastive Learning of Image Representations with Cross-Video Cycle-Consistency
• Learning Rich Representations For Structured Visual Prediction Tasks
• Self-Supervised Learning for Large-Scale Unsupervised Image Clustering
• Information based Deep Clustering: An experimental study
• New Perspective on Progressive GANs Distillation for One-class Novelty Detection
• Momentum Contrastive Autoencoder: Using Contrastive Learning for Latent Space Distribution Matching in WAE
• Semi-Supervised Self-Growing Generative Adversarial Networks for Image Recognition
• Distributed Anomaly Detection in Edge Streams using Frequency based Sketch Datastructures
• Evaluating (and improving) the correspondence between deep neural networks and human representations
• Semi-Supervised Surface Anomaly Detection of Composite Wind Turbine Blades From Drone Imagery
• Unsupervised Projection Networks for Generative Adversarial Networks
• Multimodal Image Outpainting With Regularized Normalized Diversification
• Shuffle-Then-Assemble: Learning Object-Agnostic Visual Relationship Features
• Traversing Latent Space using Decision Ferns
• Escaping from Collapsing Modes in a Constrained Space
• Force-in-domain GAN inversion
• A Survey on Adversarial Image Synthesis
• Constructing Multiple Tasks for Augmentation: Improving Neural Image Classification With K-means Features