Mode Regularized Generative Adversarial Networks

Abstract

Although Generative Adversarial Networks achieve state-of-the-art results on a variety of generative tasks, they are regarded as highly unstable and prone to miss modes. We argue that these bad behaviors of GANs are due to the very particular functional shape of the trained discriminators in high dimensional spaces, which can easily make training stuck or push probability mass in the wrong direction, towards that of higher concentration than that of the data generating distribution. We introduce several ways of regularizing the objective, which can dramatically stabilize the training of GAN models. We also show that our regularizers can help the fair distribution of probability mass across the modes of the data generating distribution, during the early phases of training and thus providing a unified solution to the missing modes problem.

Published as a conference paper at ICLR 2017

References in corpus (11)

• Conditional Generative Adversarial Nets
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling
• Energy-based Generative Adversarial Network
• Adversarial Feature Learning
• Adversarially Learned Inference
• Autoencoding beyond pixels using a learned similarity metric
• Deep multi-scale video prediction beyond mean square error
• Generating Images with Perceptual Similarity Metrics based on Deep Networks
• Temporal Generative Adversarial Nets with Singular Value Clipping
• Precomputed Real-Time Texture Synthesis with Markovian Generative Adversarial Networks

Cited by in corpus (102)

• Learning to Discover Cross-Domain Relations with Generative Adversarial Networks
• VEEGAN: Reducing Mode Collapse in GANs using Implicit Variational Learning
• How Generative Adversarial Networks and Their Variants Work: An Overview
• Generative Adversarial Networks in Computer Vision: A Survey and Taxonomy
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• Variational Approaches for Auto-Encoding Generative Adversarial Networks
• On Convergence and Stability of GANs
• An empirical study on evaluation metrics of generative adversarial networks
• A Note on the Inception Score
• Stabilizing Training of Generative Adversarial Networks through Regularization
• Maximum-Likelihood Augmented Discrete Generative Adversarial Networks
• An Introduction to Image Synthesis with Generative Adversarial Nets
• Least Squares Generative Adversarial Networks
• Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flow
• Generative Adversarial Networks (GANs) in Networking: A Comprehensive Survey & Evaluation
• Generative Adversarial Networks for Spatio-temporal Data: A Survey
• Dual Discriminator Generative Adversarial Nets
• Multi-style Generative Network for Real-time Transfer
• Adversarial Generation of Natural Language
• Generative Models for Automatic Chemical Design
• Language GANs Falling Short
• Unpaired image denoising using a generative adversarial network in X-ray CT
• Machine Learning for Physics-Informed Generation of Dispersed Multiphase Flow Using Generative Adversarial Networks
• Geometric anomaly detection in data
• Dropout-GAN: Learning from a Dynamic Ensemble of Discriminators
• Activation Maximization Generative Adversarial Nets
• An Improved Evaluation Framework for Generative Adversarial Networks
• BourGAN: Generative Networks with Metric Embeddings
• Adversarial Training in Affective Computing and Sentiment Analysis: Recent Advances and Perspectives
• Face Synthesis from Visual Attributes via Sketch using Conditional VAEs and GANs
• DiverGAN: An Efficient and Effective Single-Stage Framework for Diverse Text-to-Image Generation
• MGGAN: Solving Mode Collapse using Manifold Guided Training
• Coulomb GANs: Provably Optimal Nash Equilibria via Potential Fields
• Improved Training of Generative Adversarial Networks Using Representative Features
• Learning Generative Models across Incomparable Spaces
• Regularization Methods for Generative Adversarial Networks: An Overview of Recent Studies
• Mode Collapse and Regularity of Optimal Transportation Maps
• On the Effectiveness of Least Squares Generative Adversarial Networks
• Interactive Sketch & Fill: Multiclass Sketch-to-Image Translation
• Adversarial symmetric GANs: bridging adversarial samples and adversarial networks
• Improving Video Generation for Multi-functional Applications
• DLow: Diversifying Latent Flows for Diverse Human Motion Prediction
• Pulsar Candidate Identification with Artificial Intelligence Techniques
• A Style-Aware Content Loss for Real-time HD Style Transfer
• Distributional Adversarial Networks
• Composite Functional Gradient Learning of Generative Adversarial Models
• PI-REC: Progressive Image Reconstruction Network With Edge and Color Domain
• Inception Score, Label Smoothing, Gradient Vanishing and -log(D(x)) Alternative
• fAshIon after fashion: A Report of AI in Fashion
• Exploring AI in Fashion: A Review of Aesthetics, Personalization, Virtual Try-On, and Forecasting
• Generalization and Memorization: The Bias Potential Model
• Understanding the Effectiveness of Lipschitz-Continuity in Generative Adversarial Nets
• Recent Advances of Image Steganography with Generative Adversarial Networks
• Learning Spatial Pyramid Attentive Pooling in Image Synthesis and Image-to-Image Translation
• GDPP: Learning Diverse Generations Using Determinantal Point Process
• Training Generative Adversarial Networks via Primal-Dual Subgradient Methods: A Lagrangian Perspective on GAN
• Novelty Detection via Non-Adversarial Generative Network
• DUAL-GLOW: Conditional Flow-Based Generative Model for Modality Transfer
• Hindsight Generative Adversarial Imitation Learning
• Discriminator Feature-based Inference by Recycling the Discriminator of GANs
• Adversarially Approximated Autoencoder for Image Generation and Manipulation
• MR-GAN: Manifold Regularized Generative Adversarial Networks
• Bidirectional Generative Modeling Using Adversarial Gradient Estimation
• Generation of Synthetic Electronic Health Records Using a Federated GAN
• CRAFT: Complementary Recommendations Using Adversarial Feature Transformer
• Unpaired Multi-Domain Image Generation via Regularized Conditional GANs
• Towards Recovery of Conditional Vectors from Conditional Generative Adversarial Networks
• Deep Generative Learning via Schrödinger Bridge
• Generalized Zero-Shot Learning via VAE-Conditioned Generative Flow
• Generative Adversarial Forests for Better Conditioned Adversarial Learning
• Information Diffusion Prediction with Latent Factor Disentanglement
• Adaptive Weighted Discriminator for Training Generative Adversarial Networks
• Multimodal Image-to-Image Translation via Mutual Information Estimation and Maximization
• Reciprocal Adversarial Learning via Characteristic Functions
• A Deep Factorization of Style and Structure in Fonts
• Improving Model Robustness with Latent Distribution Locally and Globally
• Decomposed Adversarial Learned Inference
• GANs with Variational Entropy Regularizers: Applications in Mitigating the Mode-Collapse Issue
• Deconstructing Generative Adversarial Networks
• Semi-Implicit Generative Model
• Learning Shared Semantic Space with Correlation Alignment for Cross-modal Event Retrieval
• "Best-of-Many-Samples" Distribution Matching
• Paired 3D Model Generation with Conditional Generative Adversarial Networks
• Adversarial Sub-sequence for Text Generation
• Synthetic Data: Opening the data floodgates to enable faster, more directed development of machine learning methods
• Improving Text to Image Generation using Mode-seeking Function
• Post Selection Inference with Incomplete Maximum Mean Discrepancy Estimator
• IVE-GAN: Invariant Encoding Generative Adversarial Networks
• Discriminative Attribution from Counterfactuals
• Prb-GAN: A Probabilistic Framework for GAN Modelling
• A Mean-Field Theory for Kernel Alignment with Random Features in Generative and Discriminative Models
• Conditional Frechet Inception Distance
• Conditioning Trick for Training Stable GANs
• Multiclass non-Adversarial Image Synthesis, with Application to Classification from Very Small Sample
• Improving Stability of LS-GANs for Audio and Speech Signals
• Momentum Contrastive Autoencoder: Using Contrastive Learning for Latent Space Distribution Matching in WAE
• A Study of Cross-domain Generative Models applied to Cartoon Series
• Human Annotations Improve GAN Performances
• When Autonomous Systems Meet Accuracy and Transferability through AI: A Survey
• Melody-Conditioned Lyrics Generation with SeqGANs
• On the potential for open-endedness in neural networks
• Dist-GAN: An Improved GAN using Distance Constraints