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 (3)

• Conditional Generative Adversarial Nets
• Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling
• Energy-based Generative Adversarial Network

Cited by in corpus (30)

• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• Variational Approaches for Auto-Encoding Generative Adversarial Networks
• Maximum-Likelihood Augmented Discrete Generative Adversarial Networks
• Dual Discriminator Generative Adversarial Nets
• Adversarial Generation of Natural Language
• Face Synthesis from Visual Attributes via Sketch using Conditional VAEs and GANs
• Coulomb GANs: Provably Optimal Nash Equilibria via Potential Fields
• Learning Generative Models across Incomparable Spaces
• Regularization Methods for Generative Adversarial Networks: An Overview of Recent Studies
• Interactive Sketch & Fill: Multiclass Sketch-to-Image Translation
• Distributional Adversarial Networks
• PI-REC: Progressive Image Reconstruction Network With Edge and Color Domain
• Learning Spatial Pyramid Attentive Pooling in Image Synthesis and Image-to-Image Translation
• Recent Advances of Image Steganography with Generative Adversarial Networks
• Novelty Detection via Non-Adversarial Generative Network
• Hindsight Generative Adversarial Imitation Learning
• Adversarially Approximated Autoencoder for Image Generation and Manipulation
• Generalized Zero-Shot Learning via VAE-Conditioned Generative Flow
• Decomposed Adversarial Learned Inference
• GANs with Variational Entropy Regularizers: Applications in Mitigating the Mode-Collapse Issue
• Deconstructing Generative Adversarial Networks
• Learning Shared Semantic Space with Correlation Alignment for Cross-modal Event Retrieval
• "Best-of-Many-Samples" Distribution Matching
• Adversarial Sub-sequence for Text Generation
• Improving Text to Image Generation using Mode-seeking Function
• IVE-GAN: Invariant Encoding Generative Adversarial Networks
• Human Annotations Improve GAN Performances
• Improving Stability of LS-GANs for Audio and Speech Signals
• On the potential for open-endedness in neural networks
• A Study of Cross-domain Generative Models applied to Cartoon Series