f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization

Abstract

Generative neural samplers are probabilistic models that implement sampling using feedforward neural networks: they take a random input vector and produce a sample from a probability distribution defined by the network weights. These models are expressive and allow efficient computation of samples and derivatives, but cannot be used for computing likelihoods or for marginalization. The generative-adversarial training method allows to train such models through the use of an auxiliary discriminative neural network. We show that the generative-adversarial approach is a special case of an existing more general variational divergence estimation approach. We show that any f-divergence can be used for training generative neural samplers. We discuss the benefits of various choices of divergence functions on training complexity and the quality of the obtained generative models.

17 pages

References in corpus (10)

• Adam: A Method for Stochastic Optimization
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Stochastic Backpropagation and Approximate Inference in Deep Generative Models
• Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
• LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop
• Estimating divergence functionals and the likelihood ratio by convex risk minimization
• Identifying and attacking the saddle point problem in high-dimensional non-convex optimization
• A note on the evaluation of generative models
• Generative Moment Matching Networks
• How (not) to Train your Generative Model: Scheduled Sampling, Likelihood, Adversary?

Cited by in corpus (134)

• Generative Adversarial Networks: An Overview
• Generative Adversarial Network in Medical Imaging: A Review
• Improved Training of Wasserstein GANs
• NIPS 2016 Tutorial: Generative Adversarial Networks
• Deep Learning for Single Image Super-Resolution: A Brief Review
• Recent advances and clinical applications of deep learning in medical image analysis
• The Perception-Distortion Tradeoff
• Contrastive Representation Learning: A Framework and Review
• AMP: Adversarial Motion Priors for Stylized Physics-Based Character Control
• An Algorithmic Perspective on Imitation Learning
• How Generative Adversarial Networks and Their Variants Work: An Overview
• InfoVAE: Information Maximizing Variational Autoencoders
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• Privacy-preserving Spatiotemporal Scenario Generation of Renewable Energies: A Federated Deep Generative Learning Approach
• A Connection between Generative Adversarial Networks, Inverse Reinforcement Learning, and Energy-Based Models
• ASE: Large-Scale Reusable Adversarial Skill Embeddings for Physically Simulated Characters
• GANS for Sequences of Discrete Elements with the Gumbel-softmax Distribution
• Stabilizing Training of Generative Adversarial Networks through Regularization
• Multi-view Graph Contrastive Representation Learning for Drug-Drug Interaction Prediction
• Learning in Implicit Generative Models
• The Numerics of GANs
• Ten Years of Generative Adversarial Nets (GANs): A survey of the state-of-the-art
• An Introduction to Image Synthesis with Generative Adversarial Nets
• Least Squares Generative Adversarial Networks
• Loss-Sensitive Generative Adversarial Networks on Lipschitz Densities
• Generative Adversarial Networks for Spatio-temporal Data: A Survey
• Towards Deeper Understanding of Variational Autoencoding Models
• Adversarial Feature Matching for Text Generation
• Deep Knockoffs
• A survey on GANs for computer vision: Recent research, analysis and taxonomy
• Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step
• Adversarial Generation of Natural Language
• Generative Models for Automatic Chemical Design
• Sequential Recommendation with Self-Attentive Multi-Adversarial Network
• Fast and accurate simulation of particle detectors using generative adversarial networks
• Connecting Generative Adversarial Networks and Actor-Critic Methods
• Unsupervised Deep Learning Methods for Biological Image Reconstruction and Enhancement
• SCGAN: Saliency Map-guided Colorization with Generative Adversarial Network
• Learning to Draw Samples: With Application to Amortized MLE for Generative Adversarial Learning
• DisCo Fever: Robust Networks Through Distance Correlation
• Unpaired image denoising using a generative adversarial network in X-ray CT
• Stabilizing Generative Adversarial Networks: A Survey
• Maximum Entropy Generators for Energy-Based Models
• Implicit Maximum Likelihood Estimation
• Solution of Physics-based Bayesian Inverse Problems with Deep Generative Priors
• VITAL: VIsual Tracking via Adversarial Learning
• DP-GAN: Diversity-Promoting Generative Adversarial Network for Generating Informative and Diversified Text
• TIPRDC: Task-Independent Privacy-Respecting Data Crowdsourcing Framework for Deep Learning with Anonymized Intermediate Representations
• Adversarial-Residual-Coarse-Graining: Applying machine learning theory to systematic molecular coarse-graining
• Style Aggregated Network for Facial Landmark Detection
• Improved generator objectives for GANs
• Activation Maximization Generative Adversarial Nets
• Unsupervised and semi-supervised learning with Categorical Generative Adversarial Networks assisted by Wasserstein distance for dermoscopy image Classification
• PeaceGAN: A GAN-based Multi-Task Learning Method for SAR Target Image Generation with a Pose Estimator and an Auxiliary Classifier
• Understanding GANs: the LQG Setting
• A material decomposition method for dual-energy CT via dual interactive Wasserstein generative adversarial networks
• Understanding the Limitations of Variational Mutual Information Estimators
• A Tutorial on Deep Latent Variable Models of Natural Language
• The Information Autoencoding Family: A Lagrangian Perspective on Latent Variable Generative Models
• McGan: Mean and Covariance Feature Matching GAN
• GAN-QP: A Novel GAN Framework without Gradient Vanishing and Lipschitz Constraint
• Zero-shot Skeleton-based Action Recognition via Mutual Information Estimation and Maximization
• A Divergence Minimization Perspective on Imitation Learning Methods
• Global Convergence to the Equilibrium of GANs using Variational Inequalities
• Gang of GANs: Generative Adversarial Networks with Maximum Margin Ranking
• Boosting Semi-supervised Image Segmentation with Global and Local Mutual Information Regularization
• Minimax Robust Detection: Classic Results and Recent Advances
• 2-Wasserstein Approximation via Restricted Convex Potentials with Application to Improved Training for GANs
• Multinomial Adversarial Networks for Multi-Domain Text Classification
• Adversarial Message Passing For Graphical Models
• Towards a Better Understanding and Regularization of GAN Training Dynamics
• GANGs: Generative Adversarial Network Games
• Conditional Mutual Information Neural Estimator
• Some Theoretical Insights into Wasserstein GANs
• Safe Deployment for Counterfactual Learning to Rank with Exposure-Based Risk Minimization
• Age-Oriented Face Synthesis with Conditional Discriminator Pool and Adversarial Triplet Loss
• Message Passing Multi-Agent GANs
• On the Effectiveness of Least Squares Generative Adversarial Networks
• Learning Deep Energy Models: Contrastive Divergence vs. Amortized MLE
• Generative Adversarial Networks (GANs): What it can generate and What it cannot?
• On the Implicit Assumptions of GANs
• Quantitatively Evaluating GANs With Divergences Proposed for Training
• On Sampling Collaborative Filtering Datasets
• Mirror Descent Search and its Acceleration
• UGAN: Untraceable GAN for Multi-Domain Face Translation
• Variational Disentanglement for Domain Generalization
• Training Generative Adversarial Networks Via Turing Test
• Some Theoretical Properties of GANs
• Distributional Sliced-Wasserstein and Applications to Generative Modeling
• Classification Logit Two-sample Testing by Neural Networks
• Information Losses in Neural Classifiers from Sampling
• Recent Advances of Image Steganography with Generative Adversarial Networks
• Dynamics of Fourier Modes in Torus Generative Adversarial Networks
• Self-Supervised GANs with Label Augmentation
• RankGAN: A Maximum Margin Ranking GAN for Generating Faces
• New universal Lyapunov functions for non-linear reaction networks
• Learning Implicit Generative Models with the Method of Learned Moments
• Can GAN Learn Topological Features of a Graph?
• Spatial PixelCNN: Generating Images from Patches
• Cycle-Consistent Adversarial Learning as Approximate Bayesian Inference
• First Order Generative Adversarial Networks
• Variance Regularizing Adversarial Learning
• Unpaired Multi-Domain Image Generation via Regularized Conditional GANs
• Negative-Free Self-Supervised Gaussian Embedding of Graphs
• InfoFair: Information-Theoretic Intersectional Fairness
• K-Beam Minimax: Efficient Optimization for Deep Adversarial Learning
• Capturing Variabilities from Computed Tomography Images with Generative Adversarial Networks
• Correcting differences in multi-site neuroimaging data using Generative Adversarial Networks
• Filtering Variational Objectives
• Density Ratio Estimation via Infinitesimal Classification
• Hidden Convexity of Wasserstein GANs: Interpretable Generative Models with Closed-Form Solutions
• GANs with Variational Entropy Regularizers: Applications in Mitigating the Mode-Collapse Issue
• Mixability of Integral Losses: a Key to Efficient Online Aggregation of Functional and Probabilistic Forecasts
• Generative Adversarial Mapping Networks
• Featurized Density Ratio Estimation
• AMOR: Adaptive Character Control through Multi-Objective Reinforcement Learning
• Learning Segmentation Masks with the Independence Prior
• A Novel Framework for Selection of GANs for an Application
• Double Descent and Other Interpolation Phenomena in GANs
• Computing Divergences between Discrete Decomposable Models
• MI-DPG: Decomposable Parameter Generation Network Based on Mutual Information for Multi-Scenario Recommendation
• Implicit Manifold Learning on Generative Adversarial Networks
• LBS Autoencoder: Self-supervised Fitting of Articulated Meshes to Point Clouds
• On the Theoretical Equivalence of Several Trade-Off Curves Assessing Statistical Proximity
• Learning Deep Representations by Mutual Information for Person Re-identification
• Wasserstein GAN Can Perform PCA
• Restricting Greed in Training of Generative Adversarial Network
• Diversity Regularized Adversarial Learning
• Moreau-Yosida $f$-divergences
• The Representation Jensen-Rényi Divergence
• Biadversarial Variational Autoencoder
• RKL: a general, invariant Bayes solution for Neyman-Scott
• Generative Adversarial Networks with Inverse Transformation Unit
• Disentangling Latent Space for VAE by Label Relevant/Irrelevant Dimensions